Updating a column with JSON data in another table - sql

Question

I have seen a lot on JSON and SQL Server but haven't been able to find what I am looking for.
I want to update columns in one table by retrieving JSON values from another table.
Lets say I have the below table:
table : people
+-------+-----------+
| id | name |
+-------+-----------+
| 1 | John |
| 2 | Mary |
| 3 | Jeff |
| 4 | Bill |
| 5 | Bob |
+-------+-----------+
And lets pretend I have another table filled with rows of JSON like the following:
table : archive
+-------+----------------------------------------------------------------+
| id | json |
+-------+----------------------------------------------------------------+
| 1 |[{ "Column":"name","values": { "old": "Jeff", "new": "John"}}] |
| 2 |[{ "Column":"name","values": { "old": "Rose", "new": "Mary"}}] |
+-------+----------------------------------------------------------------+
Now the idea is to change Johns name to Jeff.
UPDATE people
SET name = JSON_QUERY(archive.json, '$values.old')
WHERE ID = 1
The above SQL may make no sense but I'm just trying to get across my current logic of what I'm trying to do. I hope it makes some sense.
If more information is needed please ask.

Answer 1

You can read your json using openjson and a double cross apply with with clause. Then you can use an update from to change the values in #people:
declare #people table (id int, [name] varchar(50))
insert into #people values
(1, 'John')
,(2, 'Mary')
,(3, 'Jeff')
,(4, 'Bill')
,(5, 'Bob' )
declare #json table (id int, [json] nvarchar(max))
insert into #json values
(1,'[{ "Column":"name","values": { "old": "Jeff", "new": "John"}}]')
,(2,'[{ "Column":"name","values": { "old": "Rose", "new": "Mary"}}]')
update #people
set [name] = d.old
from #people p
inner join
(
select id
, c.old
, c.new
from #json a
cross apply openjson(json) with
(
[Column] nvarchar(50)
, [values] nvarchar(MAX) as JSON
) b
cross apply openjson(b.[values]) with
(
old nvarchar(50)
, new nvarchar(50)
) c
) d
on p.id = d.id
Before update:
After update:

Answer 2

I asked you some questions in a comment above
Two remarks: Am I correct, that you mixed old and new values? And am I
correct, that the above is just a sample and you are looking for a
generic solution, where the updates might affect different columns,
maybe even more than one per row? At least the JSON would allow more
elements in the object-array.
But - as a start - you can try this:
--mockup your set-up (thanks #Andrea, I used yours)
declare #people table (id int, [name] varchar(50))
insert into #people values
(1, 'John')
,(2, 'Mary')
,(3, 'Jeff')
,(4, 'Bill')
,(5, 'Bob' )
declare #json table (id int, [json] nvarchar(max))
insert into #json values
(1,'[{ "Column":"name","values": { "old": "Jeff", "new": "John"}}]')
,(2,'[{ "Column":"name","values": { "old": "Rose", "new": "Mary"}}]')
--This will - at least - return everything you need. The rest is - assumably - dynamic statement building and EXEC():
SELECT p.*
,A.[Column]
,JSON_VALUE(A.[values],'$.old') AS OldValue
,JSON_VALUE(A.[values],'$.new') As NewValue
FROM #people p
INNER JOIN #json j ON p.id=j.id
CROSS APPLY OPENJSON(j.[json])
WITH([Column] VARCHAR(100), [values] NVARCHAR(MAX) AS JSON) A;
The result (old and new seems to be mistaken):
id name Column OldValue NewValue
1 John name Jeff John
2 Mary name Rose Mary