I am trying to build a JSON from the following table
name | flag
------+------
foo | fail
bar | pass
using the query,
DECLARE #JSONDATA nvarchar(MAX) = (SELECT [name], [flag]
FROM test
FOR JSON AUTO, ROOT('students'))
SET #JSONDATA = JSON_MODIFY(#JSONDATA, '$.class','10')
The generated JSON here is
{
"students": [
{
"name": "foo",
"flag": "fail"
},
{
"name": "bar",
"flag": "pass"
}
],
"class": "10"
}
I need to the class element at the very first node of the JSON. Is there any way, using JSON_MODIFY ?
Fiddle
At a loss forcing a sequence via modify.
Perhaps an alternative
Select class=10
,students = (SELECT [name], [flag] FROM test FOR JSON AUTO)
For JSON path, without_array_wrapper
Returns
{
"class": 10,
"students": [{
"name": "foo",
"flag": "fail"
}, {
"name": "bar",
"flag": "pass"
}]
}
EDIT- Updated SELECT as suggested by GSerg
Related
Is it possible to use JSON_MODIFY to update a dictionary value of <int,string>?
If my JSON is this:
{
"items": {
"1": "xxx",
"2": "xxx",
"3": "xxx"
}
}
I'm unable to add/update using the following:
DECLARE #Id INT = 1
DECLARE #Value NVARCHAR(50) = 'zzz'
UPDATE MyTable SET MyJsonColumn = JSON_MODIFY(MyJsonColumn, CONCAT('$.items.', #Id), #Value)
I get the error: JSON path is not properly formatted. Unexpected character '1' is found at position 10. obviously because property names can't be numeric. I can workaround the error using a <string,string> dictionary as follows:
{
"items": {
"i1": "xxx",
"i2": "xxx",
"i3": "xxx"
}
}
Then the following works fine:
DECLARE #Id INT = 1
DECLARE #Value NVARCHAR(50) = 'zzz'
UPDATE MyTable SET MyJsonColumn = JSON_MODIFY(MyJsonColumn, CONCAT('$.items.i', #Id), #Value)
However, now I'm tasked with extra work after deserializing the JSON to get the numeric ID I actually want. It's just a minor inconvenience, but I was hoping for a possible solution to keep my <int,string> dictionary. Any ideas?
I suppose I could use a dictionary like this <string,object> and simply ignore the keys when deserialized:
{
"items": {
"i1": { "id": 1, "name": "xxx" },
"i2": { "id": 1, "name": "xxx" },
"i3": { "id": 1, "name": "xxx" }
}
}
And then use
UPDATE MyTable SET MyJsonColumn = JSON_MODIFY(MyJsonColumn, CONCAT('$.items.i', #Id), JSON_QUERY(CONCAT('{"id":', #Id, ',"name":"', #Value, '"}')))
I'm also not opposed to using an array as such:
{
"items": [
{ "id": 1, "name": "xxx" },
{ "id": 2, "name": "xxx" },
{ "id": 3, "name": "xxx" }
]
}
Unfortunately, I don't know a way to update/add when in this format, and the add/update has to be done in a single UPDATE statement.
I'm hoping for some magic formula to continue using my <int,string> dictionary.
Thanks to #Charlieface, the solution is quite simple. You can navigate to numeric dictionary keys via double quotes:
DECLARE #Id INT = 1
DECLARE #Value NVARCHAR(50) = 'zzz'
UPDATE MyTable
SET MyJsonColumn = JSON_MODIFY(ISNULL(MyJsonColumn, '{"items":{}}'), CONCAT('$.items."', #Id, '"'), #Value)
How to retrieve values from employment_types (type, salary) and skills (name, level) arrays and show them in columns? I tried with employment_types and it doesn't work not to mention skills:
declare #json nvarchar(max)
set #json = '[
{
"title": "IT Admin",
"experience_level": "mid",
"employment_types": [
{
"type": "permanent",
"salary": null
}
],
"skills": [
{
"name": "Security",
"level": 3
},
{
"name": "WIFI",
"level": 3
},
{
"name": "switching",
"level": 3
}
]
},
{
"title": "Lead QA Engineer",
"experience_level": "mid",
"employment_types": [
{
"type": "permanent",
"salary": {
"from": 7000,
"to": 13000,
"currency": "pln"
}
}
],
"skills": [
{
"name": "Embedded C",
"level": 4
},
{
"name": "Quality Assurance",
"level": 4
},
{
"name": "C++",
"level": 4
}
]
}
]';
SELECT *
FROM OPENJSON(#JSON, '$.employment_types')
WITH
(
type nvarchar(50) '$.type',
salary varchar(max) '$.salary'
)
There are almost 7000 records and I'd like to show mentioned above columns from all of them.
It's hard to know exactly what you want, given that both employment_types and skills are arrays. But assuming employment_types always has only one element, you could do something like this
SELECT
j1.title,
j1.experience_level,
j1.employment_type,
salary = j1.salary_currency + ' ' + CONCAT(j1.salary_from, ' - ', j1.salary_to),
j2.name,
j2.level
FROM OPENJSON(#JSON)
WITH (
title nvarchar(100),
experience_level nvarchar(10),
employment_type nvarchar(50) '$.employment_types[0].type',
salary_from int '$.employment_types[0].salary.from',
salary_to int '$.employment_types[0].salary.to',
salary_currency char(3) '$.employment_types[0].salary.currency',
skills nvarchar(max) AS JSON
) j1
CROSS APPLY OPENJSON(j1.skills)
WITH
(
name nvarchar(50),
level int
) j2
db<>fiddle
Since we are pulling data directly from the root object, we don't need a JSON path argument. OPENJSON will automatically break out an array into separate rows. If you just wanted employment_types, you could go directly to that with a path argument.
employment_types[0] means to only get the first element of the array. If you want all the elements, you will need another OPENJSON
Note the use of AS JSON for skills, this means that the entire JSON array is pulled out, and can then be pushed through another call to OPENJSON
How to merge new array elements into an existing Json property?
Our two json objects look like this and we want to append the items of the 2nd json object to the 1st json object. (See desired result)
Is there a way to achieve this with JSON_MODIFY?
The microsoft documentation doesn't really show any example of multiple elements being merged into the already existing array. Only a single element. But we have a list of multiple elements that need to be merged.
Edit:
JSON_MODIFY(#json1, 'append $.Items', JSON_QUERY(#json2)) seems to create new array brackets instead of merging the items into the array.
Json Object 1:
DECLARE #json1 NVARCHAR(MAX) = '{
"id": 1,
"Items": [
{
"id": 1,
"name" : "Item #1"
},
{
"id": 2,
"name" : "Item #2"
}
]
}'
Json Object 2:
DECLARE #json2 NVARCHAR(MAX)='{
"Items": [
{
"id": 3,
"name": "Item #3"
},
{
"id": 4,
"name": "Item #4"
}
]
}'
Desired Result:
{
"id": 1,
"Items": [
{
"id": 1,
"name": "Item #1"
},
{
"id": 2,
"name": "Item #2"
},
{
"id": 3,
"name": "Item #3"
},
{
"id": 4,
"name": "Item #4"
}
]
}
It's Cumbersome but possible to achieve with SQL Server's built it JSON support.
First, set proper sample data (Please save us this step in your future questions):
DECLARE #Json1 nvarchar(max) =
'{
"id": 1,
"Items": [
{
"id": 1,
"name" : "Item #1"
},
{
"id": 2,
"name" : "Item #2"
}
]
}',
#Json2 nvarchar(max) =
'{
"Items": [
{
"id": 3,
"name": "Item #3"
},
{
"id": 4,
"name": "Item #4"
}
]
}';
Then, wrap a union all query containing openjson and json_query for each one of the variables with a common table expression:
With cteArray as
(
SELECT *
FROM OPENJSON(JSON_QUERY(#Json1, '$.Items'))
WITH(
Id int '$.id',
Name varchar(100) '$.name'
)
UNION ALL
SELECT *
FROM OPENJSON(JSON_QUERY(#Json2, '$.Items'))
WITH(
Id int '$.id',
Name varchar(100) '$.name'
)
)
The result of that union all query is this:
Id Name
1 Item #1
2 Item #2
3 Item #3
4 Item #4
Then, select the id from the first json using json_value, and add a subquery to select everything from the cte with for json path. Add another for json path and specify without_array_wrapper to the outer query:
SELECT JSON_VALUE(#Json1, '$.id') As id,
(
SELECT *
FROM cteArray
FOR JSON PATH
) as Items
FOR JSON PATH,
WITHOUT_ARRAY_WRAPPER
The final result:
{
"id": "1",
"Items": [{
"Id": 1,
"Name": "Item #1"
}, {
"Id": 2,
"Name": "Item #2"
}, {
"Id": 3,
"Name": "Item #3"
}, {
"Id": 4,
"Name": "Item #4"
}
]
}
You can see a live demo on Db<>Fiddle
Your problem can be solved by
SELECT dbo.udf_native_json_merge(#json1,#json2,null)
We faced similar issues trying to merge JSONs in MS SQL. We also wanted it to be recursive and allow us to define a strategy for arrays like "union", "concat" and "replace".
Our solution for JSON manipulations like merge, JSON path expressions and more is open source and available # Github
Feel free to use, comment and contribute so we can further improve JSON methods for MS SQL.
Not pretty but this will merge text and object elements. I'm sure it isn't bullet proof. Offered only as a potential solution.
Declare #json1 nvarchar(max) = '{"id":1,"messages":[{"type":"Info","text":"message1"},{"type":"Info","text":"message2"}]}'
Declare #json2 nvarchar(max) = '{"id":1,"messages":["justPlanText",{"type":"Info","text":"message3"},{"type":"Info","text":"message4"}]}'
select #json1 = case
when isjson(m.value) = 1 then
json_modify(#json1,'append $.messages',json_query(m.value))
else
json_modify(#json1,'append $.messages',m.value)
end
from openjson(json_query(#json2,'$.messages')) m
select #json1
Results for #json1:
{"id":1,"messages":[{"type":"Info","text":"message1"},{"type":"Info","text":"message2"},"justPlanText",{"type":"Info","text":"message3"},{"type":"Info","text":"message4"}]}
I'm after running into some trouble parsing a set of JSON documents using SQL Server 2017.
I've encountered a nested array (sample below) within a document which uses dynamic ObjectId's i.e. 123 as the object Key as opposed to using a static key i.e. Category and then having a separate key:value to ref the ObjectId.
As a result I can't extract the items to a table using the regular CROSS APPLY OPENJSON syntax without specifying each individual Object Id (there are thousands)?
Is there a way to do this without referencing each ObjectId explicitly ideally i'd like to return all product items in a table and just have a field withe the categoryId.
"ProductItems": {
"123": [
{
"item": "13663"
}
]
"124": [
{
"value": "2336"
},
{
"value": "3667"
}
],
"453": [
{
"value": "8667"
},
{
"value": "1956"
}
]
}
Try something like this:
DECLARE #x NVARCHAR(MAX)=
'{"ProductItems":{
"123": [
{
"item": "13663"
}
],
"124": [
{
"value": "2336"
},
{
"value": "3667"
}
],
"453": [
{
"value": "8667"
},
{
"value": "1956"
}
]
}}'
SELECT j2.*, j3.* FROM OPENJSON(#x) j1
CROSS APPLY OPENJSON(j1.Value) j2
CROSS APPLY OPENJSON(j2.Value) j3
I have the following JSON and i need to query only the name property values.
DECLARE #j NVARCHAR(4000) = N'{
"ArrayValue":[
{
"name": "XXX",
"value": 10
},
{
"name": "Memory123",
"value": 20
}
]
}'
Following is what i get with OPENJSON()
SELECT value as Name
FROM OPENJSON(#j, '$.ArrayValue')
Whcih will get the following output.
Name
---------------------------------------------------
{ "name": "Memory" }
{ "name": "Memory123" }
However I need the output in the following format.
Name
-----
Memory
Memory123
Is there any simpler way without using replace?
You can define ArrayValue as a JSON column, then use CROSS APPLY to get the name values. Here's an example:
DECLARE #j NVARCHAR(4000) = N'{
"ArrayValue":[
{
"name": "XXX",
"value": 10
},
{
"name": "Memory123",
"value": 20
}
]
}'
SELECT b.name
FROM OPENJSON(#j)
WITH (ArrayValue NVARCHAR(MAX) AS JSON) AS a
CROSS APPLY OPENJSON(a.ArrayValue)
WITH ([name] NVARCHAR(100)) AS b