MySQL 5.7如何将json数组提取到多列？

Question

I'm trying to figure out how to extract a json array to individual columns. Working in mysql 5.7.

the data i'd like to extract is held in the column 'data'. here is sample of data:

# id, member_id, event_type_id, leeds_event_id, source_event_id, staff_id, staff_name, date_start, date_end, data, created, deleted, id, source, name, created, deleted
6852184, 0034000000y3RSxAAM, 1, 1, , , Peter Nalli, , 2013-08-07 00:00:00, {"id": 1, "completed": 0, "member_id": "0034000000y3RSxAAM", "assigned_staff": "Peter Nalli", "date_completed": "2013-08-07"}, 2018-11-26 18:43:45, , 1, leeds_new.leenk_1o1s, 1 on 1s, 2018-07-24 11:14:27, 
6852185, 0034000001rmsjRAAQ, 1, 2, , , Marcus Ceniceros, , 2016-11-22 00:00:00, {"id": 2, "completed": 0, "member_id": "0034000001rmsjRAAQ", "assigned_staff": "Marcus Ceniceros", "date_completed": "2016-11-22"}, 2018-11-26 18:43:45, , 1, leeds_new.leenk_1o1s, 1 on 1s, 2018-07-24 11:14:27, 
6852186, 00340000010SqytAAC, 1, 3, , , Stephen de Man, , 2013-06-26 00:00:00, {"id": 3, "completed": 0, "member_id": "00340000010SqytAAC", "assigned_staff": "Stephen de Man", "date_completed": "2013-06-26"}, 2018-11-26 18:43:45, , 1, leeds_new.leenk_1o1s, 1 on 1s, 2018-07-24 11:14:27, 
6852187, 00340000013znbgAAA, 1, 4, , , Peter Nalli, , 2013-05-23 00:00:00, {"id": 4, "completed": 0, "member_id": "00340000013znbgAAA", "assigned_staff": "Peter Nalli", "date_completed": "2013-05-23"}, 2018-11-26 18:43:45, , 1, leeds_new.leenk_1o1s, 1 on 1s, 2018-07-24 11:14:27, 
6852188, 0034000000y3RgPAAU, 1, 5, , , Stephen de Man, , 2013-05-15 00:00:00, {"id": 5, "completed": 0, "member_id": "0034000000y3RgPAAU", "assigned_staff": "Stephen de Man", "date_completed": "2013-05-15"}, 2018-11-26 18:43:45, , 1, leeds_new.leenk_1o1s, 1 on 1s, 2018-07-24 11:14:27, 
6852189, 0034000000y3SIRAA2, 1, 6, , , Stephen de Man, , 2013-03-06 00:00:00, {"id": 6, "completed": 0, "member_id": "0034000000y3SIRAA2", "assigned_staff": "Stephen de Man", "date_completed": "2013-03-06"}, 2018-11-26 18:43:45, , 1, leeds_new.leenk_1o1s, 1 on 1s, 2018-07-24 11:14:27, 
6852190, 0034000000y3Q3DAAU, 1, 7, , , Lildella Douglas, , 2017-01-02 00:00:00, {"id": 7, "completed": 0, "member_id": "0034000000y3Q3DAAU", "assigned_staff": "Lildella Douglas", "date_completed": "2017-01-02"}, 2018-11-26 18:43:45, , 1, leeds_new.leenk_1o1s, 1 on 1s, 2018-07-24 11:14:27, 
6852191, 0034000001Cr8MjAAJ, 1, 8, , , Thomas Beer, , 2015-06-26 00:00:00, {"id": 8, "completed": 0, "member_id": "0034000001Cr8MjAAJ", "assigned_staff": "Thomas Beer", "date_completed": "2015-06-26"}, 2018-11-26 18:43:45, , 1, leeds_new.leenk_1o1s, 1 on 1s, 2018-07-24 11:14:27, 
6852192, 0034000000y3S98AAE, 1, 9, , , Peter Nalli, , 2015-03-18 00:00:00, {"id": 9, "completed": 0, "member_id": "0034000000y3S98AAE", "assigned_staff": "Peter Nalli", "date_completed": "2015-03-18"}, 2018-11-26 18:43:45, , 1, leeds_new.leenk_1o1s, 1 on 1s, 2018-07-24 11:14:27,

As an example, I'd like to take the data:

'{"id": 8, "completed": 0, "member_id": "0034000001Cr8MjAAJ", "assigned_staff": "Thomas Beer", "date_completed": "2015-06-26"}'

and extract to 5 separate new columns.

I've been trying to sort this out with json_extract() but am not getting anywhere. I'm new to this and am getting a bit lost.

Any help would be greatly appreciated.

Answer 1

In SQL, each column in the select-list must be explicit at the time you prepare the query. There's no way to make a query that expands the select-list at runtime, depending on the data it finds.

So you need to extract each field from your JSON document separately.

SELECT
  data->>'$.id' AS id,
  data->>'$.completed' AS completed,
  data->>'$.member_id' AS member_id,
  data->>'$.assigned_staff' AS assigned_staff,
  data->>'$.date_completed' AS data_completed
FROM MyTable

data->>'$.id' is equivalent to JSON_UNQUOTE(JSON_EXTRACT(data, '$.id')) .

I tested your data using MySQL 5.7.24, and this is the output of my query:

+------+-----------+--------------------+----------------+----------------+
| id   | completed | member_id          | assigned_staff | data_completed |
+------+-----------+--------------------+----------------+----------------+
| 8    | 0         | 0034000001Cr8MjAAJ | Thomas Beer    | 2015-06-26     |
+------+-----------+--------------------+----------------+----------------+

Honestly, in most cases, using JSON in MySQL would be better implemented as conventional tables and columns. When your data is only a simple set of key/value pairs, it's not worth using JSON.

---

Re comments about generated columns:

ALTER TABLE MyTable 
 ADD COLUMN completed INT AS (data->>'$.completed'),
 ADD COLUMN member_id VARCHAR(20) AS (data->>'$.member_id');

And so on. You need to add a generated column for each field you want to extract from the JSON.

Then you can query as if the fields were normal columns:

SELECT completed, member_id FROM MyTable;

+-----------+--------------------+
| completed | member_id          |
+-----------+--------------------+
|         0 | 0034000001Cr8MjAAJ |
+-----------+--------------------+

But this is not "automated" in any way.