Closed. This question needs details or clarity. It is not currently accepting answers.
Want to improve this question? Add details and clarify the problem by editing this post.
Closed 8 years ago.
Improve this question
I have a table structure like this:
Page_id || type || user_id
1 1 0
2 2 0
3 3 0
4 1 1
5 2 1
6 3 1
From this table I would like to get page_id 4,5 and 6.
But I can also have table data like this
Page_id || type || user_id
1 1 0
2 2 0
3 3 0
4 1 1
5 2 1
Then I would like to get page_id 4, 5 and 3.
So I have to get all the types, but with the priority user_id and if there is no record with user_id 1, then take the one with 0
Have tried a lot. I know I can sort it with PHP, but I hope there is a way with MySQL.
Regards Andreas
//////// ANSWER /////////
I got a lot of suggestions, and I haven't tried them all, so I can't tell it they where right or not. But I have accepted an answer, which worked for me.Thank to everybody.
SELECT a.Type, a.Page_ID
FROM table a
INNER JOIN
(SELECT Type, MAX(User_ID) AS User_ID
FROM table
GROUP BY Type ) b
ON a.Type = b.Type AND a.User_ID = b.User_ID
You can execute a SELECT query as follow
SELECT Page_id
FROM table
WHERE user_id != 0
This SQL Fiddle demonstrates the below query:
SELECT DISTINCT
(
SELECT s1.Page_id
FROM myTable AS s1
WHERE m.type = s1.type
ORDER BY s1.Page_id
LIMIT 1
) AS PageID, type,
(
SELECT s2.user_id
FROM myTable AS s2
WHERE m.type = s2.type
ORDER BY s2.Page_id
LIMIT 1
) AS User
FROM myTable AS m
The results are the records where Page_id is 1, 2, and 4. As you can see in both of the sub queries I am ordering by Page_id to make sure the data is pulled from the same record and the first Page_id for that occurrence of the type is selected.
To return only one record unique to a couple columns, you'll want to use the GROUP BY statement. Then for any other column outside of the group by columns, you need to pick an aggregate function so it knows how to summarize the value if it finds multiple records in that group. In this case you want non-zero, so max() would work
SELECT type, max(user_id) as user_id
FROM table
GROUP BY type
how about that?
select page_id,type from (
select page_id,type, user_id from mytable
group by page_id,type, user_id having user_id=max(user_id)
) as x where user_id=1
Will there ever be multiple rows for a page where user_id is not zero? Because if not (if at most you only have one row per page where user_id = 1) then this will work:
SELECT ifNull(t1.page_id,t2.page_id) as page_id, t1.type,
CASE WHEN t2.page_id IS NULL THEN t1.user_id ELSE t2.user_id END as user_id
#start with all rows (including duplicates)
FROM myTable t1
#look for a user_id > 0 for this type
LEFT OUTER JOIN myTable t2 ON t1.type = t2.type AND t2.user_id > 0
WHERE t2.page_id IS NULL # if no record with user_id > 0 found, then no need to filter
# if a record with user_id > 0 was found, then filer out the user_id = 0 record
OR (t2.page_id IS NOT NULL AND t1.user_id > 0)
See in SQLFiddle: http://sqlfiddle.com/#!2/ba877/5
Related
id parent_id child_id
1 1 1
2 2 2
3 2 2
4 1 1
I have a table from which i need to get the common values from data when i query it with id... for eg if id=2 and id=3 then return
id parent_id
2 2
3 2
i have tried this after hunting a lot through various examples :
SELECT ta.user_id,ta.interest_parent_id,ta.interest_child_id
FROM user_interest ta
WHERE ta.user_id=2 AND
(SELECT COUNT(*) FROM user_interest tb
WHERE ta.interest_parent_id=tb.interest_parent_id
AND tb.user_id=3 )>1
but it responds with only:
id parent_id
2 2
any help :( im using a mysql database with php/codeigniter to do the scripting
You can give it a try:
SELECT
tOne.id,
tOne.parent_id
FROM
(
SELECT
*
FROM user_interest A
WHERE A.id IN (2,3)
) tOne
INNER JOIN
(
SELECT
*
FROM user_interest A
WHERE A.id IN (2,3)
) tTwo
ON tOne.parent_id = tTwo.parent_id
AND tOne.id <> tTwo.id
ORDER BY tOne.parent_id;
SQL FIDDLE DEMO
Any suggestion towards optimization of the query is welcome.
EDIT: SQL FIDDLE
You can make a sub SELECT:
SELECT * FROM table WHERE Name IN (SELECT Name FROM table GROUP BY Name HAVING count(*) > 1)
Basically what i am trying to do is to suggest people based on common interests.
I have a table of Users.
I have a table of Interested_People where UserID + InterestID is stored.
I have a table of Contactlist where people who are added with each other is stored.
What I want is to only output people who are not your friends.
I searched a lot in internet but couldn't find something like so.
Although I created a query but it is very slow. Now I Kindly request you guys if you can edit my query a bit and make it much more bandwidth & time efficient.
SELECT *
FROM users
WHERE id IN(SELECT userid
FROM interested_people
WHERE interested_in IN(SELECT interested_in
FROM interested_people
WHERE userid = [userid])
AND id NOT IN(SELECT user1 AS my_friends_userid
FROM contactlist f
WHERE f.user2 = [userid]
AND accepted = 1
UNION
SELECT user2 AS my_friends_userid
FROM contactlist f
WHERE f.user1 = [userid]
AND accepted = 1))
AND id != [userid]
ORDER BY Rand ()
LIMIT 0, 10;
This query actually does the job but it takes very long about 16 sec in my local machine. and that's not what I want. I want a fast and reliable one.
Thanks in advance!
Subqueries in WHERE clauses are often slow in MySQL; at least slower than comparable JOINs.
SELECT others.*
FROM interested_people AS userI
INNER JOIN interested_people AS othersI
ON userI.interestid = othersI.interestid
AND userI.userid <> othersI.userid
INNER JOIN users AS others ON othersI.user_id = others.userid
LEFT JOIN contactlist AS cl
ON userI.userid = cl.user1
AND others.userid = cl.user2
AND cl.accepted = 1
WHERE userI.userid = [userid]
AND cl.accepted IS NULL
ORDER BY RAND()
LIMIT 0, 10;
Note: intuition makes me wonder if contactlist might be better as a where subquery.
The AND cl.accepted IS NULL ends up processed after the JOINs, resulting in allowing only results that did NOT have a match in contactlist.
If you want to enhance things a bit further:
SELECT others.*, COUNT(1) AS interestsCount
...
GROUP BY others.userid
ORDER BY interestsCount DESC, RAND()
LIMIT 0,10;
This would give you a random selection of the people that share the most interests in common.
First, looking at your interested-in query and assuming the "userID"
you are testing with is = 1. Sounds like you are trying to get one level
away from those user 1 is also interested in...
SELECT userid FROM interested_people
WHERE interested_in IN
( SELECT interested_in FROM interested_people
WHERE userid = [userid] )
Sample Data for Interested_People
userID Interested_In
1 5
1 7
1 8
2 3
2 5
2 7
7 1
7 2
7 5
8 3
In this case, the innermost returns interested_in values of 5, 7, 8.
Then, getting all users who are interested in 5, 7 and 8 would return 2 and 7.
(but since both users 2 and 7 are interested in 5, the 2 ID would be returned TWICE
thus a possible duplicate join later on. I would do distinct. This same
result could be done with the following query which you could sample times with...
SELECT distinct ip2.userid
from
interested_people ip
join interested_people ip2
ON ip.interested_in = ip2.interested_in
where
userid = [parmUserID]
Now, you need to exclude from this list all your contacts already accepted.
You could then left-join TWO TIMES for the from/to contact and ensure NULL
indicating not one of the contacts... Then join again to user table to
get the user details.
SELECT
u.*
from
users u
JOIN
( SELECT distinct
ip2.userid
from
interested_people ip
join interested_people ip2
ON ip.interested_in = ip2.interested_in
left join contactList cl1
ON ip2.userid = cl1.user1
AND cl1.accepted = 1
left join contactList cl2
ON ip2.userid = cl2.user2
AND cl2.accepted = 1
where
ip.userid = [parmUserID]
AND NOT ip2.userID = [parmUserID] ) PreQuery
ON u.id = PreQuery.userID
order by
RAND()
limit
0, 10
I would have two indexes on your contactList table to optimize both left-joins... with user1 and user2 in primary position... Similarly for the interested_people table.
table index
contactList ( user1, accepted )
contactList ( user2, accepted )
interested_people ( userid, interested_in )
interested_people ( interested_in, userid )
I would expect your user table is already indexed on the ID as primary key.
I think this will give you the same results but perform a lot better:
SELECT * FROM Users u
INNER JOIN interested_people i
ON u.id = i.userid
WHERE NOT EXISTS
(SELECT * FROM contacts WHERE user1 = [userid] or user2 = [userid] and accepted=1)
AND id != [userid]
ORDER BY Rand()
LIMIT 0, 10
Skip the ORDER BY clause if that is at all reasonable. That will be the most expensive part
The select and join clauses give you the users who are interested in connecting and the WHERE NOT EXISTS is a performant way to exclude those contacts already listed.
One of my table has a field user_ids and the value of the field like 2,3
group_id| user_ids
--------|------------
1 | 2,3
--------|------------
2 | 5,8
I want to update the field without deleting the current value. For ex. If I need to add 5 for group_id id 1, then 2,3 should be like 2,3,5
I m using this query:
UPDATE users_group SET user_ids = CONCAT( SUBSTRING( user_ids, 1, CHAR_LENGTH( user_ids ) -1 ) , ',5' ) WHERE group_id =1
But it is deleting previous value with comma.
group_id| user_ids
--------|------------
1 | ,5
--------|------------
2 | 5,8
can anyone suggest the right way for this?
Can you not just concatenate it on, rather than trying to split it up first?
UPDATE users_group
SET user_ids = CONCAT_WS(',', user_ids, '5' )
WHERE group_id =1
But this does suggest a badly normalised database design. Generally a comma separated list should instead be stored as rows on another table (ie, one row per value in the list) as suggested by Mark Baker.
EDIT - If you want to only have a single copy of any id in each user_ids field, irrespective of how many times you try to insert it, and you want to be able to add multiple ids at once:-
UPDATE users_group a
INNER JOIN
(
SELECT 3 AS an_id
UNION
SELECT 4
) b
ON FIND_IN_SET(b.an_id, a.user_ids) = 0
SET a.user_ids = CONCAT_WS(',', a.user_ids, b.an_id )
WHERE a.group_id =1
EDIT again - if you have a table of users containing the ids then you can select the ids from that where the id is one of those you want to add.
Something like this.
UPDATE users_group a
INNER JOIN
(
SELECT id
FROM users
WHERE id IN (3, 4)
) b
ON FIND_IN_SET(b.id, a.user_ids) = 0
SET a.user_ids = CONCAT_WS(',', a.user_ids, b.id )
WHERE a.group_id =1
update table1 set name = concat(name, ', ', 5) WHERE group_id =1
Please try this query. It may be useful for you.
UPDATE users_group SET user_ids = CONCAT( user_ids , ',5' ) WHERE group_id =1
Try the below query:
UPDATE users_group
SET user_ids = CONCAT( user_ids , ',5' )
WHERE group_id =1
I have a php web page and I got a problem with counting data using group by in mysql query
When any question submitted, id is automatically add and idqu=0
if question replied then idqu is equal to replied question's id
My table and data
id idqu question answer user date
1 0 quest1 test 28042014
2 0 quest2 scott 29042014
3 2 reply1 andy 01052014
4 0 quest3 test 01052014
5 4 reply2 scot 01052014
My question is how can I count question's reply?
quest1 (0)
quest2 (1)
quest3 (1)
This query does the job. I gave the table the name "tbl"
Select q.id, count(a.id) as answers
From tbl q
Left join tbl a on a.idqu=q.id
Where q.idqu=0
Group by q.id
Try this
Select t1.question, count(t2.id) as counts
From table1 t1
Left join table1 t2 on t2.idqu=t1.id
Where t1.idqu=0
Group by t1.id
DEMO
select idqu, count(idqu) as reply_count
from table_name
where idqu <> 0
group by idqu
This is for every replies, with a bit of php. You don't need group by I think.
$query="SELECT * From YourTable WHERE idqu>0";
$res=mysql_query($query);
mysql_num_rows($res);
If you want to select replies on one question, add an AND idqu=(yourInt) in the query.
Hope I did understand and replied correctly!
If I understand correctly, try:
SELECT question, SUM(CASE WHEN idqu > 0 THEN 1 ELSE 0 END)
FROM tablename
GROUP BY question
I have a query which give result like
id | productid | userid | coinsid
1 | 2 | 2 | 5
3 | 2 | 2 | 6
4 | 2 | 3 | 7
5 | 2 | 4 | 8
6 | 2 | 3 | 9
This is result for specific productid. Now i have to update the balance in user table by adding $1 to all the users in above result, but if userid is twice, i need to add $1 twice to the balance of that specific user. So in the above case $1 twice added to userid=2 balance and userid=3 balance.
The simple way is to count records for every distinct userid and run queries as many time as we have users in foreach loop. But i am looking for some optimize way. Please suggest any. Thanks
One approach:
UPDATE user_table u
JOIN ( SELECT q.userid
, SUM(1.00) AS deposit
FROM (
-- original OP query goes here
) q
GROUP BY q.userid
) r
ON r.userid = u.userid
SET u.balance = u.balance + r.deposit
We use the original OP query that returns the resultset displayed, and make that an inline view (aliased in the query above as q).
From that, we query a distinct list of userid, and the number of times that userid appears in the resultset. That gives us the username and a deposit amount (1 dollar for each time the userid appears) (some databases might want us to specify the value as 1.0 rather than 1, to make sure it was decimal. I think the SUM is more representative of what we are trying to accomplish.)
We join that inline view (r) to the user table, and add the deposit amount to the current balance, for that user (assuming the balance is stored as decimal dollars (1.00 = one dollar)
To testing, convert the UPDATE into a SELECT statement:
remove the "SET" clause
add an "ORDER BY" clause (optional) to make the results determinate
remove the "UPDATE" keyword and replace it
with:
SELECT r.userid
, r.deposit
, u.balance AS old_balance
, u.balance + r.deposit AS new_balance
, u.userid
FROM
Full select:
SELECT r.userid
, r.deposit
, u.balance AS old_balance
, u.balance + r.deposit AS new_balance
, u.userid
FROM user_table u
JOIN ( SELECT q.userid
, SUM(1.00) AS deposit
FROM (
-- original OP query goes here
) q
GROUP BY q.userid
) r
ON r.userid = u.userid
NOTE There is no WHERE clause, the JOIN predicates (in the ON clause) is what determines which rows are selected/affected in the user table.
Assuming you have no duplicate user ids in your balance table, maybe something like this would work:
update balance_table set balance_table.balance = (select count(*) from users_table where users_table.user_id = balance_table.user_id) * 1;
I haven't tried this query against a mysql database as I am more familiar with plsql, but wouldn't something like this work ?
The correlated subquery in the other answer will work, but an INNER JOIN will usually be more efficient. Try something like this; you'll of course need to supply the table and column names.
UPDATE myTable
INNER JOIN (
SELECT userid, count(*) AS AmountToAdd
FROM users
GROUP BY userid
) UserCounts ON myTable.userid = UserCounts.userid
SET balance = balance + UserCounts.AmountToAdd
select count(*), userid from yourTable group by userid
If I do understand your question.