Update a nested value from a Json field - mysql

Question

Consider this table:
DROP TABLE IF EXISTS `example`;
CREATE TABLE `example` (
`id` int NOT NULL AUTO_INCREMENT,
`content` json NOT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb3;
And these rows:
INSERT INTO example(content)
VALUES (
'[ { "date": "1617210148", "name": "John", "status": "0" },
{ "date": "1617210148", "name": "Jack", "status": "0" },
{ "date": "1617210148", "name": "Henry", "status": "0" }]'
);
I'd like to update the value of the key status where name = Jack to 1
The result would be:
{ "date": "1617210148", "name": "Jack", "status": "1" }
How can I do this using JSON_REPLACE() or JSON_SET() in a SQL query (I'm aiming for a partial update since I'm using MySQL 8.0.25)?

Answer 1

This is very awkward, nearly impossible with MySQL's JSON functions.
You can use JSON_REPLACE() or JSON_SET(), but both require that you know the path to the field you want to change. So in this case, we can see that the array element is $[1] but if you didn't know that, you couldn't use this solution.
mysql> select json_pretty(json_replace(content, '$[1].status', '1')) as j
from example\G
*************************** 1. row ***************************
j: [
{
"date": "1617210148",
"name": "John",
"status": "0"
},
{
"date": "1617210148",
"name": "Jack",
"status": "1"
},
{
"date": "1617210148",
"name": "Henry",
"status": "0"
}
]
The question like yours has come up before on Stack Overflow, for example JSON update single value in MySQL table. The solution in that case depends on you knowing which array element your pseudo-record exists in.
You can get the path to a JSON element with JSON_SEARCH(), but you can only search by value, not by a key/value pair. If "Jack" occurs in some other field, that would also be found.
mysql> select json_unquote(json_search(content, 'one', 'Jack')) as path from example;
+-----------+
| path |
+-----------+
| $[1].name |
+-----------+
To search for a key/value pair, you need to use JSON_TABLE() and that requires you upgrade to MySQL 8.0. And that doesn't tell you the path to the element, it only allows you to return a specific row out of the array.
mysql> select j.* from example cross join json_table(content, '$[*]' columns(
date int unsigned path '$.date',
name varchar(10) path '$.name',
status int path '$.status')
) as j where name = 'Jack';
+------------+------+--------+
| date | name | status |
+------------+------+--------+
| 1617210148 | Jack | 0 |
+------------+------+--------+
Here's a trick: You can extract the name field, and that turns into an array of those values:
mysql> select json_extract(content, '$[*].name') as a from example;
+---------------------------+
| a |
+---------------------------+
| ["John", "Jack", "Henry"] |
+---------------------------+
Then you can search that array to get the array position:
mysql> select json_search(json_extract(content, '$[*].name'), 'one', 'Jack') as root from example;
+--------+
| root |
+--------+
| "$[1]" |
+--------+
Some unquoting and adding .status and you can get the full path to the field you want to update:
mysql> select concat(json_unquote(json_search(json_extract(content, '$[*].name'), 'one', 'Jack')), '.status') as path from example;
+-------------+
| path |
+-------------+
| $[1].status |
+-------------+
Now use it in a JSON_SET() call:
mysql> select json_pretty(
json_set(content,
concat(json_unquote(json_search(json_extract(content, '$[*].name'), 'one', 'Jack')), '.status'),
'1')) as newcontent
from example\G
*************************** 1. row ***************************
newcontent: [
{
"date": "1617210148",
"name": "John",
"status": "0"
},
{
"date": "1617210148",
"name": "Jack",
"status": "1"
},
{
"date": "1617210148",
"name": "Henry",
"status": "0"
}
]
Using this in an UPDATE looks like this:
mysql> update example set content = json_set(content, concat(json_unquote(json_search(json_extract(content, '$[*].name'), 'one', 'Jack')), '.status'), '1');
That's a long way to go. Now compare how difficult that is to:
UPDATE content SET status = 1 WHERE name = 'Jack';
Storing data in a JSON document when you eventually want to use SQL expressions to search or update individual fields within the JSON document is a costly mistake. It increases the complexity of any code you write to do it, and the developer who needs to take over maintenance of your code after you have moved on to another project will curse your name.