SQL query with join to get nested array of objects - mysql

Question

Summary: I'll start with a JSON schema to describe the expectation. Notice the roles with a nested array of objects and I'm looking for a "Smart query" that can fetch it one single query.
{
"id": 1,
"first": "John",
"roles": [ // Expectation -> array of objects
{
"id": 1,
"name": "admin"
},
{
"id": 2,
"name": "accounts"
}
]
}
user
+----+-------+
| id | first |
+----+-------+
| 1 | John |
| 2 | Jane |
+----+-------+
role
+----+----------+
| id | name |
+----+----------+
| 1 | admin |
| 2 | accounts |
| 3 | sales |
+----+----------+
user_role
+---------+---------+
| user_id | role_id |
+---------+---------+
| 1 | 1 |
| 1 | 2 |
| 2 | 2 |
| 2 | 3 |
+---------+---------+
Attempt 01
In a naive approach I'd run two sql queries in my nodejs code, with the help of multipleStatements:true in connection string. Info.
User.getUser = function(id) {
const sql = "SELECT id, first FROM user WHERE id = ?; \
SELECT role_id AS id, role.name from user_role \
INNER JOIN role ON user_role.role_id = role.id WHERE user_id = ?";
db.query(sql, [id, id], function(error, result){
const data = result[0][0]; // first query result
data.roles = result[1]; // second query result, join in code.
console.log(data);
});
};
Problem: Above code produces the expected JSON schema but it takes two queries, I was able to narrow it down in a smallest possible unit of code because of multiple statements but I don't have such luxury in other languages like Java or maybe C# for instance, there I've to create two functions and two sql queries. so I'm looking for a single query solution.
Attempt 02
In an earlier attempt With the help of SO community, I was able to get close to the following using single query but it can only help to produce the array of string (not array of objects).
User.getUser = function(id) {
const sql = "SELECT user.id, user.first, GROUP_CONCAT(role.name) AS roles FROM user \
INNER JOIN user_role ON user.id = user_role.user_id \
INNER JOIN role ON user_role.role_id = role.id \
WHERE user.id = ? \
GROUP BY user.id";
db.query(sql, id, function (error, result) {
const data = {
id: result[0].id, first: result[0].first,
roles: result[0].roles.split(",") // manual split to create array
};
console.log(data);
});
};
Attempt 02 Result
{
"id": 1,
"first": "John",
"roles": [ // array of string
"admin",
"accounts"
]
}
it's such a common requirement to produce array of objects so wondering there must be something in SQL that I'm not aware of. Is there a way to better achieve this with the help of an optimum query.
Or let me know that there's no such solution, this is it and this is how it's done in production code out there with two queries.
Attempt 03
use role.id instead of role.name in GROUP_CONCAT(role.id), that way you can get hold of some id's and then use another subquery to get associated role names, just thinking...
SQL (doesn't work but just to throw something out there for some thought)
SELECT
user.id, user.first,
GROUP_CONCAT(role.id) AS role_ids,
(SELECT id, name FROM role WHERE id IN role_ids) AS roles
FROM user
INNER JOIN user_role ON user.id = user_role.user_id
INNER JOIN role ON user_role.role_id = role.id
WHERE user.id = 1
GROUP BY user.id;
Edit
Based on Amit's answer, I've learned that there's such solution in SQL Server using JSON AUTO. Yes this is something I'm looking for in MySQL.
To articulate precisely.
When you join tables, columns in the first table are generated as
properties of the root object. Columns in the second table are
generated as properties of a nested object.

Answer 1

User this Join Query
FOR JSON AUTO will return JSON for your query result
SELECT U.UserID, U.Name, Roles.RoleID, Roles.RoleName
FROM [dbo].[User] as U
INNER JOIN [dbo].UserRole as UR ON UR.UserID=U.UserID
INNER JOIN [dbo].RoleMaster as Roles ON Roles.RoleID=UR.RoleMasterID
FOR JSON AUTO
out put of above query is
[
{
"UserID": 1,
"Name": "XYZ",
"Roles": [
{
"RoleID": 1,
"RoleName": "Admin"
}
]
},
{
"UserID": 2,
"Name": "PQR",
"Roles": [
{
"RoleID": 1,
"RoleName": "Admin"
},
{
"RoleID": 2,
"RoleName": "User"
}
]
},
{
"UserID": 3,
"Name": "ABC",
"Roles": [
{
"RoleID": 1,
"RoleName": "Admin"
}
]
}
]

Answer 2

Though it is an old question, just thought might help others looking for the same issue. The below script should output the json schema you have been looking for.
SELECT roles, user.* from `user_table` AS user
INNER JOIN `roles_table` AS roles
ON user.id=roles.id