Compile/combine the contents of several records.

[imani_technology_spam]

I have the following table;

CREATE TABLE [x_Note] (
[x_NoteId] [int] IDENTITY (1, 1) NOT NULL ,
[Note] [varchar] (7200) COLLATE SQL_Latin1_Gene ral_Pref_CP1_CI _AS NOT
NULL ,

CONSTRAINT [PK_x_NoteId] PRIMARY KEY CLUSTERED
(
[x_NoteId],
) WITH FILLFACTOR = 90 ON [USERDATA] ,

) ON [USERDATA]
GO

My clients want me to take the contents of the Note column for each row
and combine them. In other words, they basically want:

Note = Note [accumulated from previous rows] + Char(13) [because they
want a carriage return] + Note [from current record].

What is the most efficient and relatively painless way to do this? I
think it might require a cursor, but I'm not sure if there is a more
elegant set-based method to make this happen.

[Stu]

There's not a set-based method; the BEST method is to use the client
application to accomplish this. Visual Basic, C#, Java, etc, are all
desigined for munging arrays in this fashion.

If you must do this on the SQL side (for email reports, etc), then you
can use a cursor. But cursor performance will suck.

Stu

[imani_technology_spam]

Which kind of cursor would "suck" the least? Is there a way to
construct the code to be more efficient?

[Stu]

SQL Server is not really designed to handle cursors; in some cases,
they are unavoidable, but they just don't perform well compared to a
set-based solution. What client are you using to present this data to
your customers? Whatever it is has to be better than a SQL cursor.

Stu

[imani_technology_spam]

Trust me, this has to be a cursor. I'm just wondering how I can make
it happen.

[David Portas]

im************* ******@yahoo.co m wrote:

I have the following table;

CREATE TABLE [x_Note] (
[x_NoteId] [int] IDENTITY (1, 1) NOT NULL ,
[Note] [varchar] (7200) COLLATE SQL_Latin1_Gene ral_Pref_CP1_CI _AS NOT
NULL ,

CONSTRAINT [PK_x_NoteId] PRIMARY KEY CLUSTERED
(
[x_NoteId],
) WITH FILLFACTOR = 90 ON [USERDATA] ,

) ON [USERDATA]
GO

My clients want me to take the contents of the Note column for each row
and combine them. In other words, they basically want:

Note = Note [accumulated from previous rows] + Char(13) [because they
want a carriage return] + Note [from current record].

What is the most efficient and relatively painless way to do this? I
think it might require a cursor, but I'm not sure if there is a more
elegant set-based method to make this happen.

This looks pretty odd. If the average size of data in the Note column
is just half of your maximum size then you'll only be able to
concatenate two rows before you break the 8000 character ceiling. This
means you will most likely have to return a result set rather than a
variable (TEXT variables not allowed). But you want each value
delimited with carriage returns and most clients will display a
multiple row result set as multiple lines anyway - so what can you hope
to gain from concatenating them?

The above comments apply to SQL Server 2000. In SQL Server 2005 you can
do some fancy stuff like the following. Please always specify what
version you are using so that we don't have to guess.

WITH X (note, row_no)
AS
(
SELECT CAST(note AS VARCHAR(MAX)),
ROW_NUMBER() OVER (ORDER BY x_noteid)
FROM x_note
)
SELECT
MAX(CASE row_no WHEN 1 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE row_no WHEN 2 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE row_no WHEN 3 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE row_no WHEN 4 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE row_no WHEN 5 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE row_no WHEN 6 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE row_no WHEN 7 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE row_no WHEN 8 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE row_no WHEN 9 THEN note+CHAR(13) ELSE '' END)
/* ... etc */
AS note
FROM X ;

SELECT REPLACE( REPLACE(
(SELECT note
FROM x_note
ORDER BY
x_noteid
FOR XML PATH (''))
, '<note>', CHAR(13)),'</note>','') AS note ;

IMO clientside is the best option though.

--
David Portas
SQL Server MVP
--

[Erland Sommarskog]

im************* ******@yahoo.co m (im************ *******@yahoo.c om) writes:

I have the following table;

CREATE TABLE [x_Note] (
[x_NoteId] [int] IDENTITY (1, 1) NOT NULL ,
[Note] [varchar] (7200) COLLATE SQL_Latin1_Gene ral_Pref_CP1_CI _AS
NOT
NULL ,

CONSTRAINT [PK_x_NoteId] PRIMARY KEY CLUSTERED
(
[x_NoteId],
) WITH FILLFACTOR = 90 ON [USERDATA] ,

) ON [USERDATA]
GO

My clients want me to take the contents of the Note column for each row
and combine them. In other words, they basically want:

Note = Note [accumulated from previous rows] + Char(13) [because they
want a carriage return] + Note [from current record].

What is the most efficient and relatively painless way to do this? I
think it might require a cursor, but I'm not sure if there is a more
elegant set-based method to make this happen.

In SQL 2000, a cursor is the only defined way to do this, but you
can only compose a string which is 8000 chars long.

In SQL 2005, you can do this painlessly, thanks to some syntax from
the XML corner of SQL 2005:

select substring(List, 1, datalength(List )/2 - 1)
-- strip the last CR from the list
from
(select Note + char(13) as [text()]
from x_Note
order by x_NoteId
for xml path('')) as Dummy(List)

A bit obscure, but it works!
--
Erland Sommarskog, SQL Server MVP, es****@sommarsk og.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

[imani_technology_spam]

Okay, now I have a new variation to this problem. Let's change the DDL
slightly:

CREATE TABLE [x_Note] (
[x_NoteId] [int] IDENTITY (1, 1) NOT NULL ,
[NoteCategory] [int] NOT NULL ,
[Note] [varchar] (7200) COLLATE
SQL_Latin1_Gene ral_Pref_CP1_CI _AS NOT
NULL ,
CONSTRAINT [PK_x_NoteId] PRIMARY KEY CLUSTERED
(
[x_NoteId],
) WITH FILLFACTOR = 90 ON [USERDATA] ,
) ON [USERDATA]
GO

Instead of combining the contents of the Note column for each row, we
need to combine the Notes within each NoteCategory. In other words, I
need to combine the Notes for all rows of NoteCategory #1, all the
Notes for NoteCategory #2, and so on.

In other words, FOR EACH NoteCategory, the clients want:

Note = Note [accumulated from previous rows within the NoteCategory] +
Char(13) [because they want a carriage return] + Note [from current
record within the NoteCategory].

Now, I have NO idea how to pull this off. I was considering nested
cursors, but I'm not sure if that is the best way, or even a practical
way. Help!!!

[David Portas]

im************* ******@yahoo.co m wrote:

Okay, now I have a new variation to this problem. Let's change the DDL
slightly:

CREATE TABLE [x_Note] (
[x_NoteId] [int] IDENTITY (1, 1) NOT NULL ,
[NoteCategory] [int] NOT NULL ,
[Note] [varchar] (7200) COLLATE
SQL_Latin1_Gene ral_Pref_CP1_CI _AS NOT
NULL ,
CONSTRAINT [PK_x_NoteId] PRIMARY KEY CLUSTERED
(
[x_NoteId],
) WITH FILLFACTOR = 90 ON [USERDATA] ,
) ON [USERDATA]
GO

Instead of combining the contents of the Note column for each row, we
need to combine the Notes within each NoteCategory. In other words, I
need to combine the Notes for all rows of NoteCategory #1, all the
Notes for NoteCategory #2, and so on.

In other words, FOR EACH NoteCategory, the clients want:

Note = Note [accumulated from previous rows within the NoteCategory] +
Char(13) [because they want a carriage return] + Note [from current
record within the NoteCategory].

Now, I have NO idea how to pull this off. I was considering nested
cursors, but I'm not sure if that is the best way, or even a practical
way. Help!!!

You still didn't tell us what version of SQL Server you are using. Nor
have you explained why you can't do this client side.

Try:

SELECT notecategory,
MAX(CASE seq WHEN 1 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE seq WHEN 2 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE seq WHEN 3 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE seq WHEN 4 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE seq WHEN 5 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE seq WHEN 6 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE seq WHEN 7 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE seq WHEN 8 THEN note+CHAR(13) ELSE '' END)+
MAX(CASE seq WHEN 9 THEN note+CHAR(13) ELSE '' END)
/* ... etc */
AS note
FROM
(SELECT T1.notecategory , T1.x_noteid, T1.note, COUNT(*) seq
FROM x_note AS T1
JOIN x_note AS T2
ON T1.notecategory = T2.notecategory
AND T1.x_noteid >= T2.x_noteid
GROUP BY T1.notecategory , T1.x_noteid, T1.note) AS T
GROUP BY notecategory ;

--
David Portas
SQL Server MVP
--

[imani_technology_spam]

Sorry. I am using SQL Server 2000, and we can't do this client side
because: 1) we are doing a data migration from an older app with a SQL
Server 2000 back end and 2) our clients won't let us take a client-side
approach to this problem.