I'm searching for an equivalent for MySQL "FIELD_IN_SET()" with joined tables. Here is an example of what I want to do.
I have three tables. The first contains fields "user_id" and "user_name". The second contains fields "id", "user_id" and "role_id". The third contains "id", "role_name".
I want to search for users that owns two different specified roles. After LEFT JOINs, my table will look like this :
user_id - user_name - role_name
12 - Yoda - jedi
12 - Yoda - master
15 - Obi-Wan Kenobi - jedi
The first question is : How to search for users that are both "jedi" and "master".
The second question is : Is it better (optimization || readability) to do it on PHP side (getting all results in an array and do some loop to verify what I want to do) or directly in the MySQL query, knowing that I will have to build the query dynamically (it's for a PHP/MySQL search engine in a great user database).
Thank you :) .
user1527491
This problem is called RELATION DIVISION
SELECT a.user_name
FROM firstTable a
INNER JOIN secondTable b
ON a.user_ID = b.user_ID
INNER JOIN thirstTable c
ON b.role_ID = c.role_ID
WHERE c.role_NAME IN ('jedi', 'master')
GROUP BY a.user_name
HAVING COUNT(*) = 2
SQL of Relational Division
If a unique constraint was not define for role_NAME for every user_name, DISTINCT keyword is needed.
SELECT a.user_name
FROM firstTable a
INNER JOIN secondTable b
ON a.user_ID = b.user_ID
INNER JOIN thirstTable c
ON b.role_ID = c.role_ID
WHERE c.role_NAME IN ('jedi', 'master')
GROUP BY a.user_name
HAVING COUNT(DISTINCT c.role_NAME) = 2
Related
Hi I am trying to change some code so that users with certain group id do not show in search results.
at the moment the code is
$query = 'SELECT distinct b.'.$db->nameQuote('id')
.' FROM '.$db->nameQuote('#__users').' b';
I am trying to add something like the following but cannot get it to work.
select * from '.$db->quoteName('#__users').' where id not in (select user_id from #__user_usergroup_map where group_id = 8)
Any help would be appreciated.
Thanks.
This how you can try
select * from users u
where
not exists
(
select 1
from user_group ug
where
u.id = ug.iduser
AND ug.group_id = 8
)
You need change the table name and field name in the above query as your need.
DEMO
There are three ways of doing that.
Method 1: Join
Join usergroup_map table and then, select all rows that have different group_id that way:
SELECT <comma separated fields> from USERS u
LEFT JOIN USERGROUPS ug ON u.user_id = ug.user_id
WHERE ug.group_id != :group_id
GROUP BY u.user_id
And then -> bind any number to :group_id.
Please keep in mind that above statement will return users that are assigned to :group_id only if they are in another groups as well. (one to many relation).
Method 2: Subselect
Already suggested by other users.
Method 3: Join & Subselect
The 3rd, and the last option is to use sub-query within join statement, which tends to be the fastest solution in terms of optimalization. Of course, query times may be different depending on the engine you are working with, but it's worth giving a try.
SELECT <comma separated fields>
FROM USERS u
INNER JOIN (
SELECT g.user_id
FROM USERGROUPS g
WHERE g.group_id != :group_id
) ug ON ug.user_id = u.user_id
--
In this answer, I assume that you have properly configured database with foreign keys.
I am working with four tables:
query,
store,
cluster_group,
tv_region
I want to retrieve query records belonging to a particular TV Region record.
A query record either belongs to a record in the store or cluster_group tables. The query table has 'store_id' and 'cluster_group_id' columns. Either will be null whilst the other will refer to a record in the store or cluster_group table. The store and cluster_group tables both have a 'tv_region_id' column.
To retrieve query records that belong to a TV Region record that has an id = 2, I wrote the following SQL statement:
SELECT query.id AS query_id, cluster_group.name as cluster_name, cluster_group.tv_region_id as cluster_tv_region, store.store_name
FROM query
INNER JOIN cluster_group
ON cluster_group.id=query.cluster_group_id
INNER JOIN store
ON store.id=query.store_id
WHERE cluster_group.tv_region_id = 2
AND store.tv_region_id = 2;
The problem is that it returns zero records even though there are query records that belong to the specified TV Region (via a cluster group or store record). I may have misunderstood how 'inner join' works.
I'm working with Doctrine 2 and have attempted the query using the left join but it still returns nothing, I guess this is because its returning null values too. How do I get it to return only the query records I am interested in and no null values?
Appreciate if someone can point me in the right direction to retrieve the relevant query records.
You will need to use OUTER joins, or possibly LEFT
INNER joins will only return data where there is a matching record on both sides. As you state in your question you only get a match on one side or the other, therefore there is never going to be a matching record across both store and cluster
Jeff Atwood has a good explanation of the join types on his blog http://www.codinghorror.com/blog/2007/10/a-visual-explanation-of-sql-joins.html
Try this:-
SELECT query.id AS query_id, cluster_group.name as cluster_name, cluster_group.tv_region_id as cluster_tv_region, store.store_name
FROM query
LEFT OUTER JOIN cluster_group
ON cluster_group.id=query.cluster_group_id
AND cluster_group.tv_region_id = 2
LEFT OUTER JOIN store
ON store.id=query.store_id
AND store.tv_region_id = 2
EDIT - no experience of doctrine, but you could try using a UNION:-
SELECT query.id AS query_id, cluster_group.name as cluster_name, cluster_group.tv_region_id as cluster_tv_region, NULL AS store_name
FROM query
INNER JOIN cluster_group
ON cluster_group.id=query.cluster_group_id
AND cluster_group.tv_region_id = 2
UNION
SELECT query.id AS query_id, NULL as cluster_name, NULL as cluster_tv_region, store.store_name
FROM query
INNER JOIN store
ON store.id=query.store_id
AND store.tv_region_id = 2
It works like this for me:
from table1 inner join table2 on table1.id = table2.id
inner join table3 on table1.id = table3.id
I have the following table structure of my db:
tbl_project
tbl_employee
tbl_deliverable
user_to_deliverable
Where tbl_prjct and tbl_deliverable have a 1-to-many relationship,
tbl_employee and tbl_deliverable have many-many relationships so they are split into user_to_deliverable table.
I want a query to show project_name(from tbl_project), project's deliverables (from tbl_deliverable) and employee name which that specific deliverable is assigned to.
Can I get help writing this query?
Your desired query is much like 89.67% like this :D
SELECT a.ProjectName,
b.deliverables,
d.employeename
FROM tbl_project a
INNER JOIN tbl_deliverable b
ON a.projectID = b.projectID
INNER JOIN user_to_deliverable c
ON b.recordID = c.RecordID
-- or could be the primary key
INNER JOIN tbl_employee d
ON c.userID = d.userID
Approximate solution without knowing your schema. If you want the deliverables/employees concatenated for each project it would probably be best done by retrieving the results and parsing in a non-SQL language (PHP, Python, or whatever you are using). If you want this completely done in SQL you will need to use GROUP_CONCAT().
select tbl_project.name as project,
tbl_deliverable.name as deliverable,
tbl_employee.name as employee
from tbl_project
join tbl_deliverable
on (tbl_project.id = tbl_deliverable.project_id)
join user_to_deliverable
on (tbl_deliverable.id = user_to_deliverable.deliverable_id)
join tbl_employee
on (user_to_deliverable.employee_id = tbl_employee.id)
Suppose that in table "ab" I have the names of the students that get along from class "a" and class "b", identically I have table "ac" and "bc". What SQL query should I use in order to get all the combinations possible of students who can form groups (i.e. "get along together")? And how can i extend this to n classes?
For example: John from class a gets along with Jen from class b and Steff from class c, and Jen and Steff get along. Therefore John, Jen and Steff can form a group).
For this I would create two tables, a student table (id, name, class) and a relationship table (student1, student2). You might also want to add a class table for the time, location etc of the class.
A friendship would have two relationships (2,3) and (3,2) to describe it as two way. One way might be a follower or fan of another student.
This will scale up to a lot more than 3 classes.
Then you can use multiple joins to get friends of friends and so on to an arbitrary depth.
Here is a query to get friends of friends (fof):
SELECT fof_details.*
FROM relationships r
INNER JOIN relationships fof
ON r.student2 = fof.student1
INNER JOIN student fof_details
ON fof_details.id = fof.student2
WHERE r.student1 = '12';
There are also database engines made specifically for doing graph modeling like this.
http://openquery.com/blog/graph-engine-mkii
This query should return all students who can be in one group with John.
WITH ABC AS (SELECT AB.A, AB.B, AC.C FROM (SELECT * FROM AB
INNER JOIN BC
ON AB.B=BC.B)
INNER JOIN AC
ON (AC.C=BC.C AND AB.A=AC.A))
SELECT STUDENT FROM (
SELECT AB.B STUDENT FROM ABC WHERE AB.A='John'
UNION
SELECT AC.C STUDENT FROM ABC WHERE AB.A='John')
GROUP BY STUDENT
PS.: Written fast without any syntax check, hope you'll be able to bring this to work :)
The initial query can be satisfied by the code
select ab.a, ab.b, ac.c
from
ab inner join
bc on ab.b = bc.b inner join
ac on ac.a = ab.a and bc.c = ac.c
Stepping up to n classes will get progressively more complex as n=4 would be the same query with the additional three joins
inner join ad on ab.a = ad.a
inner join bd on bd.b = ab.b and ad.d = bd.d
inner join cd on cd.c = ac.c and ad.d = cd.d
2 classes requires 1 table and no joins,
3 classes requires 3 tables and 2 joins,
4 classes requires 6 tables and 5 joins
So we can see it getting progressively more complex as we proceed
First you don't want to have a table for each class. You are capturing the same type of information in multiple tables and this is generally considered a bad practice. You want to "normalize" your data so that the same data exists in one place.
Second, name your tables appropriately so that you understand what you are actually trying to build. Maybe you are generalizing to mask what your intentions for the actual implementations are by using "ab" in the question, but if you are doing this in your actual code it will hurt you in the long run.
It appears you need a people table with names and a friends table where you track who is friends with who:
create table people ( id int, name char(128) );
create table friends ( id int, person_id int, friend_id int );
Then you just need to have the query to get the groups:
SELECT person.* FROM friends
INNER JOIN friends grp
ON friends.friend_id = grp.person_id
INNER JOIN people person
ON person.id = grp.friend_id
WHERE friends.person_id = 42;
So, I understand how the relationships work in mysql but I'm having a hard time figuring out how its implemented in my code.
For example, say I have the 3 tables.
Table 1: users - user id, username, user city
Table 2: categories - category id, category name
Table 3: user_categories - user id, category id
If I were to query the database for every user that was in a particular city and list them out with the all of the categories they belong to... How would I do this? Would I need to loop through the results and do a separate query for each user, then list the results? Or, is there some magic query that will return a multidimensional array?
I believe the above would be many-to-many, correct me if I'm wrong....
EDIT In the user_categories table, a user can contain more than 1 category, I'm trying to figure out how to return all of them
Thanks!
You're absolutely right, it is a many-to-many query.
And from what I understand, what you're looking for is the ability to have some kind of hierarchical result to display, meaning for one user, have an array of all the categories he's assigned to...
Couple of things you could do:
Option 1: Query the users table:
SELECT u.user_id, u.username, u.user_city WHERE city = 'somecity';
From the results, get all the user_id's that match, put them in an array.
array(1,3,4,5)
Then execute a query by joining the 2 tables categories and user_categories, and passing the array as a comma separated list in a where in:
SELECT user_categories.user_id, categories.category_name
FROM user_categories INNER JOIN categories ON user_categories.category_id = categories.category_id
WHERE user_categories.user_id IN (1,3,4,5)
This will give you a list of user-id, category name that you can use in your script with the previous results to build your result set
option 2: my preferred, use MySQL's GROUP_CONCAT(http://dev.mysql.com/doc/refman/5.0/en/group-by-functions.html#function_group-concat).
SELECT users.user_id, users.user_name, GROUP_CONCAT(categories.category_name) AS categories
FROM users
INNER JOIN user_categories ON users.id = users_categories.user_id
INNER JOIN categories ON user_categories.category_id = category.id
WHERE user.user_city = 'somecity'
GROUP BY user.user_id
This will return something like:
user_id username categories
1 u1 cat1, cat2, cat3
2 u2 cat1, cat3
You can specify the separator by using SEPARATOR in group_concat.
You need to JOIN the tables.
If I were to query the database for every user that was in a particular city and list them out with the all of the categories they belong to
SELECT *
FROM users
INNER JOIN user_categories
ON (user_id)
INNER JOIN categories
ON (category_id)
WHERE ...
You could try:
SELECT u.user_id, u.username, u.user_city, c.category_id, c.category_name
FROM users u
INNER JOIN user_categories uc ON u.user_id = uc.user_id
INNER JOIN categories c ON uc.category_id = c.category_id
WHERE u.user_city = 'Cityname';
I haven't tested this, and there might be a more efficient way to do it, but it should work.
If you are unfamiliar with joins in mysql, check this out.