How can i select multiple row and cascade them? - sql

Question

Let's say I have three tables:
table1 fields:
memberid | name
table2 fields:
interestId | interestName
table3 (used to make a relation between member and interest) fields:
memberid | interestId
and now I know I can user inner join to select one member's all interests.
But how can I cascade all the interests in a single row???
For example, I can select this result:
memberid name interstId interestName
1 dennis 1 play basketball
1 dennis 2 music
1 dennis 3 moive
but the result i want to get is:
memberid name interests
1 dennis play basketball, music, moive
How can I write the SQL query?
Thanks in advance!

Answer 1

In SQL Server 2005 onwards, You can use XML Path() to concatenate values. It appears to be very performant too.
EDIT : Have tested the following and works
SELECT
t1.memberid,
t1.[name],
ISNULL(STUFF(
(
SELECT
', ' + t2.interestName
FROM
table2 t2
INNER JOIN
table3 t3
ON
t2.interestId = t3.interestId
WHERE
t3.memberid = t1.memberid
FOR XML PATH('')
), 1, 2, ''
), 'None') As interests
FROM
table1 t1
GROUP BY
t1.memberid,
t1.[name]
Example code:
DECLARE #table1 TABLE ( memberid INT IDENTITY(1,1), name VARCHAR(25) )
INSERT INTO #table1 VALUES('dennis');
INSERT INTO #table1 VALUES('mary');
INSERT INTO #table1 VALUES('bill');
DECLARE #table2 TABLE ( interestId INT IDENTITY(1,1), interestName VARCHAR(25) )
INSERT INTO #table2 VALUES('play basketball');
INSERT INTO #table2 VALUES('music');
INSERT INTO #table2 VALUES('movie');
INSERT INTO #table2 VALUES('play hockey');
INSERT INTO #table2 VALUES('wine tasting');
INSERT INTO #table2 VALUES('cheese rolling');
DECLARE #table3 TABLE ( memberid INT, interestId INT )
INSERT INTO #table3 VALUES(1,1);
INSERT INTO #table3 VALUES(1,2);
INSERT INTO #table3 VALUES(1,3);
INSERT INTO #table3 VALUES(2,2);
INSERT INTO #table3 VALUES(2,4);
INSERT INTO #table3 VALUES(2,6);
INSERT INTO #table3 VALUES(3,1);
INSERT INTO #table3 VALUES(3,5);
INSERT INTO #table3 VALUES(3,6);
SELECT
t1.memberid,
t1.[name],
ISNULL(STUFF(
(
SELECT
', ' + t2.interestName
FROM
#table2 t2
INNER JOIN
#table3 t3
ON
t2.interestId = t3.interestId
WHERE
t3.memberid = t1.memberid
FOR XML PATH('')
), 1, 2, ''
), 'None') As interests
FROM
#table1 t1
GROUP BY
t1.memberid,
t1.[name]
Results
memberid name interests
----------- -----------------------------------------------------------------------
1 dennis play basketball, music, movie
2 mary music, play hockey, cheese rolling
3 bill play basketball, wine tasting, cheese rolling

Answer 2

It depends on the DB you are using. Take a look at this question: Show a one to many relationship as 2 columns - 1 unique row (ID & comma separated list)

Answer 3

since you didn't specify your database, I can advice to take a look at left (right) joins.

Answer 4

http://dev.mysql.com/doc/refman/5.0/en/group-by-functions.html#function_group-concat
GROUP_CONCAT
just saw the comment on the dbms system, but it would work with mysql.

Answer 5

SELECT t1.memberid, t1.name,
STUFF(
( SELECT ', ' + interestName
FROM table2 t2
inner join table3 as t3
on t2.interestId = t3.interestId and t3.memberid = t1.memberid
FOR XML PATH('') //use to merge the interests
), 1, 2, ''
) As interests
FROM table1
This works

Answer 6

Depends on particular database. Maybe it will help you (using T-SQL and MS SQL Server) for a known memberid:
declare #result varchar(8000)
set #result = ''
declare #memberid int
set #memberid = 1
select #result = str(#memberid) + ' ' + (select name from table1 where memberid = #memberid) + ' '
select #result = #result + str(interestid) + ' ' + interest
from
(
select table2.interestid, table2.interestname
from table3
inner join table2 on table2.interestid = table3.interestid
where table3.memberid = #memberid
) t1
select left(#result, LEN(#result) - 1)