I'm required to supply a json object
like this:
[
{
id: '59E59BC82852A1A5881C082D5FFCAC10',
user: {
...users[1],
last_message: "16-06-2022",
topic: "Shipment"
},
unread: 2,
},
{
id: '521A754B2BD028B13950CB08CDA49075',
user: {
...users[2],
last_message: "15-06-2022",
topic: "Settings"
},
unread: 0,
}
]
it is not difficult for me to build a json
like this:
(with this fiddle https://dbfiddle.uk/?rdbms=sqlserver_2019&fiddle=bf62626de20d3ca7191aa9c1ef0cd39b)
[
{
"id": "59E59BC82852A1A5881C082D5FFCAC10",
"user": {
"id": 1,
"last_message": "16-06-2022",
"topic": "Shipment"
},
"unread": 2
},
{
"id": "521A754B2BD028B13950CB08CDA49075",
"user": {
"id": 2,
"last_message": "15-06-2022",
"topic": "Settings"
},
"unread": 1
},
{
"id": "898BB874D0CBBB1EFBBE56D2626DC847",
"user": {
"id": 3,
"last_message": "18-06-2022",
"topic": "Account"
},
"unread": 1
}
]
but I have no idea how to put the ...users[1]
, instead of "id": 1
into user
node:
is there a way?
CodePudding user response:
This is not actually valid JSON, but you can create it yourself using STRING_AGG
and CONCAT
SELECT
'[' STRING_AGG(u.spread, ',') ']'
FROM (
SELECT
spread = CONCAT(
'{id:''',
u.userId,
''',user:{...users[',
ROW_NUMBER() OVER (ORDER BY u.id),
'],last_message: "',
t.created_at,
'",topic:"',
t.topic,
'"},unread:',
(SELECT COUNT(*) FROM @tickets t3 WHERE t3.userId = u.userId AND t3.read_at IS NULL),
'}'
)
FROM @Users u
CROSS APPLY (
SELECT top 1
t.ticketId,
t.created_at,
t.topic
FROM @Tickets t
WHERE t.userId = u.userId
ORDER BY
t.created_at DESC
) t
) u
Note that you may need to escape values, but I don't know how this not-JSON works so couldn't say.