How to select JSON Properties as column using T-SQL - json

Question

I have a table with a JSON column. I want to select JSON properties as column. The property names will be unknown. So I have to use dynamic SQL. Based on this SO suggestion, I was able to get properties.
CREATE TABLE [Templates]
(
[ID] [INT] NOT NULL,
[Template] [NVARCHAR](MAX)
)
INSERT INTO Templates(ID,Template)
VALUES (1, '{"FirName":"foo"}'),
(2, '{"FirName":"joe","LastName":"dow"}'),
(3, '{"LastName":"smith","Address":"1234 Test Drive"}'),
(4, '{"City":"New York"}')
// SELECT Keys
SELECT DISTINCT(j.[key])
FROM Templates T
CROSS APPLY OPENJSON(T.Template) AS j
How do I create fitting statement/WITH-clause dynamically to select properties as column? If property doesn't exist then it should return null
SQL FIDDLE

Answer 1

Another possible approach is to use OPENJSON() with dynamically generated WITH clause. Note, that in this case you need to use lax mode in the path expression to guarantee that OPENJSON() doesn't raise an error if the object or value on the specified path can't be found.
Table:
CREATE TABLE [Templates](
[ID] [int] NOT NULL,
[Template] [nvarchar](max)
)
INSERT INTO Templates(ID,Template)
VALUES
(1,'{"FirName":"foo"}'),
(2,'{"FirName":"joe","LastName":"dow"}'),
(3,'{"LastName":"smith","Address":"1234 Test Drive"}'),
(4,'{"City":"New York"}')
Statement:
DECLARE #stm nvarchar(max) = N''
-- Dynamic explicit schema (WITH clause)
SELECT #stm = CONCAT(
#stm,
N', [',
[key],
N'] nvarchar(max) ''lax $."',
[key],
'"'''
)
FROM (
SELECT DISTINCT j.[key] FROM Templates t
CROSS APPLY OPENJSON(T.Template) AS j
) cte
-- Statement
SELECT #stm = CONCAT(
N'SELECT j.* ',
N'FROM Templates t ',
N'CROSS APPLY OPENJSON(t.Template) WITH (',
STUFF(#stm, 1, 2, N''),
N') j '
)
-- Execution
PRINT #stm
EXEC sp_executesql #stm
Output:
--------------------------------------------
Address City FirName LastName
--------------------------------------------
foo
joe dow
1234 Test Drive smith
New York

Answer 2

Dynamic columns would require Dynamic SQL. If the desired columns are known, you can use a simple pivot or even a conditional aggregation.
Example
Declare #SQL varchar(max)= stuff((Select ','+QuoteName([key])
From (SELECT DISTINCT(j.[key]) FROM Templates T
CROSS APPLY OPENJSON(T.Template) AS j) A
Order By 1
For XML Path('')),1,1,'')
Set #SQL = '
Select *
From (
Select T.ID
,j.[Key]
,j.[Value]
From Templates T
Cross Apply OpenJSON(T.Template) AS j
) src
Pivot ( max(value) for [Key] in ('+ #SQL+') ) pvt
'
Exec(#SQL)
EDIT - If you don't want ID in the Final Results
Declare #SQL varchar(max)= stuff((Select ','+QuoteName([key])
From (SELECT DISTINCT(j.[key]) FROM Templates T
CROSS APPLY OPENJSON(T.Template) AS j) A
Order By 1
For XML Path('')),1,1,'')
Set #SQL = '
Select '+#SQL+'
From (
Select T.ID
,j.[Key]
,j.[Value]
From Templates T
Cross Apply OpenJSON(T.Template) AS j
) src
Pivot ( max(value) for [Key] in ('+ #SQL+') ) pvt
'
Exec(#SQL)