SELECT DISTINCT u.id AS userId,u.type AS userType
FROM User AS u,Personal AS p,Company AS c
WHERE (p.realName LIKE '%adf%' AND u.type=1 AND u.id=p.userId)
OR (c.name LIKE '%grge%' AND u.id=c.userId)
LIMIT 0 , 10000
You can write your query as:
SELECT DISTINCT u.id AS userId,u.type AS userType
FROM User AS u inner join Personal AS p on u.id=p.userId
inner join Company AS c on u.id=c.userId
where p.realName LIKE '%adf%' or c.name LIKE '%grge%'
LIMIT 0 , 10000
Try to avoid comma seperated JOINS
You appear to be doing a quite hideous cross join, and then selectively narrowing down the records in the WHERE clause.
It is probably better to do 2 queries and union the results together. Each query can do one proper join. It is still going to have to access one column using the LIKE, and with a leading wild card that is not going to be quick (it can't use indexes).
SELECT u.id AS userId,
u.type AS userType
FROM User AS u
INNER JOIN Personal AS p
ON u.id = p.userId
WHERE p.realName LIKE '%adf%'
AND u.type = 1
UNION
SELECT u.id AS userId,
u.type AS userType
FROM User AS u
INNER JOIN Company AS c
ON u.id=c.userId
WHERE c.name LIKE '%grge%'
LIMIT 0 , 10000
Related
I have a huge performance issue with my below query I figured that what is the leakage but don't know how to go around it.
the issue is that inner selects go through all the records in tables which is like 200K and then try to select and apply the filter to it so the whole data get selected in the first round.
SELECT *
FROM (
SELECT
( SELECT GROUP_CONCAT(tp.login) login
FROM tp
WHERE tp.user_id = ue.user_id ) login,
u.email as email,
ue.fname as name
FROM user_extra ue
LEFT JOIN users u ON u.id = ue.user_id
) t
WHERE
email like '%sradesign.net#gmail.com%'
OR fname like '%test%'
OR login like '%461988%
Test does I have not made a mistake:
SELECT GROUP_CONCAT(tp.login) login,
u.email as email,
ue.fname as name
FROM user_extra ue
JOIN users u ON u.id = ue.user_id -- I doubt in LEFT
JOIN tp ON tp.user_id = ue.user_id
GROUP BY ue.fname,
u.email
-- , ue.user_id -- maybe needed, but I doubt
HAVING email like '%sradesign.net#gmail.com%'
OR fname like '%test%'
OR login like '%461988%
This is intermediate variant - it must be divided on two (or maybe three - depends on your data) separate queries combined with UNION later.
sometimes it happens that one user doesn't have any login(there is no record for him in tp table) but I want to show him if the customer search term matched in the email of fname how to do this? – Ali Mahmoudi
In such case this table must be joined using outer joining (LEFT JOIN). – Akina
with left join again become like before performance issue the query run for 200 Second – Ali Mahmoudi
As I have said above - this is intermediate variant. If it is correct then let's divide taking into account possible absence in tp:
SELECT GROUP_CONCAT(tp.login) login,
u.email as email,
ue.fname as name
FROM user_extra ue
JOIN users u ON u.id = ue.user_id -- I doubt in LEFT
LEFT JOIN tp ON tp.user_id = ue.user_id
WHERE u.email like '%sradesign.net#gmail.com%'
GROUP BY ue.fname,
u.email
UNION ALL
SELECT GROUP_CONCAT(tp.login) login,
u.email as email,
ue.fname as name
FROM user_extra ue
JOIN users u ON u.id = ue.user_id -- I doubt in LEFT
LEFT JOIN tp ON tp.user_id = ue.user_id
WHERE ue.fname like '%test%'
GROUP BY ue.fname,
u.email
UNION ALL
SELECT GROUP_CONCAT(tp.login) login,
u.email as email,
ue.fname as name
FROM user_extra ue
JOIN users u ON u.id = ue.user_id -- I doubt in LEFT
JOIN tp ON tp.user_id = ue.user_id
GROUP BY ue.fname,
u.email
HAVING login like '%461988%
Test its output for correctness.
If it will produce duplicates then replace UNION ALL with UNION DISTINCT.
You seem to have an extra layer of SELECTs. This causes an extra temp table that contains a huge amount of data.
Unnecessary use of LEFT adds confusion and may prevent some optimizations.
Leading wildcards prevent usage of index.
OR is hard to optimize. But...
A common trick for optimizing OR:
( SELECT ... WHERE email like '%sradesign.net#gmail.com%' )
UNION DISTINCT
( SELECT ... WHERE fname like '%test%' )
UNION DISTINCT
( SELECT ... WHERE login like '%461988% )
However, because of the JOINs, UNION may not be practical.
Consider using FULLTEXT instead of LIKE.
I have two tables: users and lessons.
Currently I select all the users using:
SELECT * FROM users WHERE user_type = 1;
Then use PHP to loop through them and count their number of lessons using
SELECT COUNT(*) FROM lessons WHERE student_id=users.user_id;
I would like to combine this into a single query and I'm pretty sure this is possible with a JOIN but it is beyond my basic SQL knowledge.
Do this all in one query. If you want a count per user:
select u.user_id, count(l.student_id)
from users u left join
lessons l
on u.user_id = l.student_id
where u.user_type = 1
group by u.user_id
You can use a join, or a correlated subquery:
select
u.*,
(select count(*) from lessons l where l.student_id = u.user_id) no_lessons
from users u
The upside of the subquery solution is that it does not require aggregation in the outer query. With an index on lessons(student_id), this should be an efficient option.
You can write the following query for this:
SELECT * FROM users u LEFT JOIN lessons l ON u.user_id=l.student_id WHERE u.user_type=1 GROUP BY u.user_id
This will join both the tables (users and lessons) based on the id of both tables and the GROUP BY clause will group all the records of same id as you just want the number of lessons per user.
I am presently in a DBS course and I am working on an inline view:
SELECT userId, firstname, lastname, gender
FROM
(SELECT COUNT(dvdId) dvdId, u.userId
FROM userDVD
JOIN users u ON userDVD.userId = u.userId
GROUP BY userId) as T
WHERE gender = 'F';
When I run the Query it returns the error unknown column in field list. If I try to specify
u.firstname, u.lastname, u.gender
I return the same error. Any thoughts?
SELECT T.userId, T.firstname, T.lastname, T.gender
FROM (
SELECT users.userId, users.firstname, users.lastname, users.gender
FROM userDVD
JOIN users ON userDVD.userId = users.userId
WHERE gender = 'F' GROUP BY userId
) as T;
I worked through it and it turns out I didn't realize that because I had to alias my inline view that I needed to specify it in the original select statement. This concept was sadly not covered in my course. I appreciate those who gave helpful tips versus rude comments. So, Thank you Drew and Matt
Drew is right that you have a derived table T, but I'll attempt to add some details which might be useful to you as you are in a course.
you have a derived table T
an outer query (first select)
an inner query (second select in brackets)
the inner query uses a table alias of u for users.
the final columns of the inner query are dvdId which is a count of the dvdIds in userDvd table and userId
Because an outer query is constrained to use the final recordset of the inner query only those 2 columns are available to be selected and seeing they don't include firstname, lastname, or gender you are recieving the error.
If you want to keep your query the same but to use those columns you could join your derived table back to the user table and get the columns you desire. Such as this:
SELECT userId, firstname, lastname, gender
FROM
(SELECT COUNT(dvdId) dvdId, u.userId
FROM userDVD
JOIN users u ON userDVD.userId = u.userId
GROUP BY userId) as T
INNER JOIN users u2
ON t.user_id = u2.user_id
WHERE gender = 'F';
That technique to join back to the original table is great for when you have to optimize an aggregation that is on a large table or look for duplicates or something like that. However in your case you really just need a single query aggregating on all of the columns you want in the result set such as this:
SELECT
u.userId, u.firstname, u.lastname, u.gender, COUNT(d.dvdId) as DVDCount
FROM
userDVD d
JOIN users u
ON d.userId = u.userId
WHERE u.gender = 'F'
GROUP BY
u.userId, u.firstname, u.lastname, u.gender
;
I'm trying to write a script that will export modx users to CSV, fairly straightforward stuff, but in modx users can belong to many groups. Simply joining the modx_member_groups table will result in several rows for different users.
What I would like to do is somehow rewrite the query below so that the join on the modx_member_groups would return a list or array of group ids that the user belongs to.
For example, I would like the returned data to look like:
user_group | id | username | ...the rest
1,3,5,7 | 12 | johndoe | ...
here is the query I have.
SELECT mg.user_group, u.id, u.username, ua.*
FROM modx_users u
LEFT JOIN modx_user_attributes ua ON u.id = ua.internalKey
LEFT JOIN modx_member_groups mg ON u.id = mg.member
LIMIT 10
Ideally it would be awesome to somehow select the actual group names as columns. and then just force a true or false in the group name column.
UPDATE
I've updated the query after shtever's answer but have performance issues:
- GROUP_CONCAT was returning a BLOB type so I had to convert it, setting the group_concat_max_length to below 512 was not working
SELECT GROUP_CONCAT(CONVERT(mg.user_group, CHAR(10)) ORDER BY mg.user_group SEPARATOR ',') AS groups, u.id, u.username, ua.*
FROM modx_users u
LEFT JOIN modx_user_attributes ua ON u.id = ua.internalKey
LEFT JOIN modx_member_groups mg ON mg.member = u.id
GROUP BY u.id
The query now takes 27.5 seconds to execute if I limit it to 10 results or let it run on the entire 6000 users it always takes 27.5 seconds. If I remove the GROUP_CONCAT ~ same amount of time.
For MySQL, take a look at the GROUP_CONCAT function. Mysql GROUP_CONCAT Description
Your query might look something like:
SELECT GROUP_CONCAT(DISTINCT mg.user_group ASC SEPARATOR ',') , u.id, u.username, ua.*
FROM modx_users u
LEFT JOIN modx_user_attributes ua ON u.id = ua.internalKey
LEFT JOIN modx_member_groups mg ON u.id = mg.member
GROUP BY u.id, u.username
LIMIT 10
You might have to fiddle with the GROUP BY fields depending on the relation between the modx_user_attributes and modx_users table.
The following query does what I want. It returns all the resuls in the users table and then if there is a match in the details tble, returns the relevant data
users
id|username
details
id|userid|firstname|lastname
$sql = "SELECT u.*, d.*
FROM `users` u
LEFT JOIN `details` d on
u.id = d.userid
ORDER BY $strorder";
However, when I try to join an additonal table where I want to do the same thing--return all the results of the users table and if there is a match in the third table, return the relevant data (total followers of this user)--it only returns one record.
3rd table
follow
id|followerid|followedid
$sql = "SELECT u.*, d.*, COUNT(f.id)
FROM `users` u
LEFT JOIN `details` d on
u.id = d.userid
LEFT JOIN `follow` f on
u.id = f.followedid
ORDER BY $strorder";
Can anyone see what I am doing wrong? Thanks in advance for any suggestions.
Many thanks.
Try to avoid * to select fields, it will be clearer to group your datas (even if mysql is quite permissive with groupings).
When you have an aggregate function (like COUNT, SUM), the other "non aggregated" requested fields should be in a GROUP BY clause.
Mysql don't force you to GROUP BY all the fields, but... I think it's quite a good habit to be "as ANSI as possible" (usefull when you use another DBMS)
SELECT u.id, u.username, d.firstname, d.lastname, count(*) as numberfollowers
FROM user u
LEFT JOIN details d on u.id = d.userid
LEFT JOIN follow f on u.id = f.followedid
GROUP BY u.id, u.username, d.firstname, d.lastname --or just GROUP BY u.id with Mysql
ORDER BY count(*) desc
COUNT being an aggregate function, when selected with other columns, requires you to group your results by those other columns in the select list.
You should rewrite your query with columns that you want to select from users and details and group by those columns.