How to Insert JSON into a Table in SQL Server

Posted on by Ian

If you have a JSON document that you need to insert into a table in a SQL Server database, the OPENJSON() function could be just what you need.

OPENJSON() is a table-valued function that returns JSON in tabular format. That is, it converts your JSON into a tabular result set consisting of rows and columns. Therefore, it enables you to insert it into a table.

Example 1 – SELECT INTO

In this example, we use SELECT * INTO to create a new table and insert the contents of the JSON document into it.

DECLARE @json NVARCHAR(4000) = N'{ 
    "pets" : {
            "cats" : [
            { "id" : 1, "name" : "Fluffy", "sex" : "Female" },
            { "id" : 2, "name" : "Long Tail", "sex" : "Female" },
            { "id" : 3, "name" : "Scratch", "sex" : "Male" }
        ],
            "dogs" : [
            { "name" : "Fetch", "sex" : "Male" },
            { "name" : "Fluffy", "sex" : "Male" },
            { "name" : "Wag", "sex" : "Female" }
        ]
    }
}';

SELECT * INTO JsonCats1
FROM OPENJSON(@json, '$.pets.cats')
WITH  (
        CatId     int             '$.id',  
        CatName   varchar(60)     '$.name', 
        Sex       varchar(6)      '$.sex', 
        Cats      nvarchar(max)   '$' AS JSON   
    );

I first declared a variable and put the JSON into it. Then I used a SELECT * INTO statement to insert its contents.

However, you’ll notice that I used a WITH clause to define a schema. Basically what I’m doing here is creating my own column names and their respective data types, then mapping each JSON key with a column.

On the last line, I use AS JSON to specify that the contents of that column is a JSON object or array.

This will become clear when I select the contents of the table.

Let’s do that.

SELECT * FROM JsonCats1;

Result:

+---------+-----------+--------+------------------------------------------------------+
| CatId   | CatName   | Sex    | Cats                                                 |
|---------+-----------+--------+------------------------------------------------------|
| 1       | Fluffy    | Female | { "id" : 1, "name" : "Fluffy", "sex" : "Female" }    |
| 2       | Long Tail | Female | { "id" : 2, "name" : "Long Tail", "sex" : "Female" } |
| 3       | Scratch   | Male   | { "id" : 3, "name" : "Scratch", "sex" : "Male" }     |
+---------+-----------+--------+------------------------------------------------------+

So we can see that the first three columns each contain a different value from the JSON document, and the last column contains the actual JSON for each array element.

We can also use the sys.column system catalog view check the table’s column names and types.

SELECT
    name AS [Column],
    TYPE_NAME(system_type_id) AS [Type],
    max_length
FROM sys.columns 
WHERE OBJECT_ID('JsonCats2') = object_id;

Result:

+----------+----------+--------------+
| Column   | Type     | max_length   |
|----------+----------+--------------|
| Cat Id   | int      | 4            |
| Cat Name | varchar  | 60           |
| Sex      | varchar  | 6            |
| Cats     | nvarchar | -1           |
+----------+----------+--------------+

Again, exactly how we’d specified it.

Note that sys.columns always returns a max_length of -1 when the column data type is varchar(max), nvarchar(max), varbinary(max), or xml. We specified nvarchar(max) and so the value of -1 is exactly as expected.

Also note that, when you use AS JSON (as we did on the fourth column), you must make that column an nvarchar(max).

Example 2 – INSERT INTO

Here’s the same example, except this time we insert the JSON into a table that already exists.

Therefore, the first thing we need to do is create the table:

CREATE TABLE [dbo].[JsonCats2](
	[CatId] [int] NULL,
	[CatName] [varchar](60) NULL,
	[Sex] [varchar](6) NULL,
	[Cats] [nvarchar](max) NULL
) ON [PRIMARY] TEXTIMAGE_ON [PRIMARY]
GO

Now that we’ve created that, we can go ahead and insert the contents of our JSON document into that table.

Like this:

DECLARE @json NVARCHAR(4000) = N'{ 
    "pets" : {
            "cats" : [
            { "id" : 1, "name" : "Fluffy", "sex" : "Female" },
            { "id" : 2, "name" : "Long Tail", "sex" : "Female" },
            { "id" : 3, "name" : "Scratch", "sex" : "Male" }
        ],
            "dogs" : [
            { "name" : "Fetch", "sex" : "Male" },
            { "name" : "Fluffy", "sex" : "Male" },
            { "name" : "Wag", "sex" : "Female" }
        ]
    }
}';

INSERT INTO JsonCats2
SELECT * 
FROM OPENJSON(@json, '$.pets.cats')
WITH  (
        CatId     int             '$.id',  
        CatName   varchar(60)     '$.name', 
        Sex       varchar(6)      '$.sex', 
        Cats      nvarchar(max)   '$' AS JSON   
    );

The only difference between this and the previous example is that I replaced the following bit:

SELECT * INTO JsonCats1

With this:

INSERT INTO JsonCats2
SELECT *

So, selecting the contents of the table will produce the same result as the previous example.

SELECT * FROM JsonCats2;

Result:

+---------+-----------+--------+------------------------------------------------------+
| CatId   | CatName   | Sex    | Cats                                                 |
|---------+-----------+--------+------------------------------------------------------|
| 1       | Fluffy    | Female | { "id" : 1, "name" : "Fluffy", "sex" : "Female" }    |
| 2       | Long Tail | Female | { "id" : 2, "name" : "Long Tail", "sex" : "Female" } |
| 3       | Scratch   | Male   | { "id" : 3, "name" : "Scratch", "sex" : "Male" }     |
+---------+-----------+--------+------------------------------------------------------+

Example 3 – Using the Default Schema

In the previous examples I defined my own schema. That is, I specified the names of the columns for the tables, and I specified the actual data types of those columns.

If I hadn’t done that, OPENJSON() would have used the default schema. The default schema consists of three columns; key, value, and type.

Here’s an example of using the default schema when inserting JSON into a table.

DECLARE @json NVARCHAR(4000) = N'{ 
    "pets" : {
            "cats" : [
            { "id" : 1, "name" : "Fluffy", "sex" : "Female" },
            { "id" : 2, "name" : "Long Tail", "sex" : "Female" },
            { "id" : 3, "name" : "Scratch", "sex" : "Male" }
        ],
            "dogs" : [
            { "name" : "Fetch", "sex" : "Male" },
            { "name" : "Fluffy", "sex" : "Male" },
            { "name" : "Wag", "sex" : "Female" }
        ]
    }
}';

SELECT * INTO JsonCats3
FROM OPENJSON(@json, '$.pets.cats');

So the only difference between this and the first example, is that I removed the whole WITH clause. That’s the bit that defined the schema in the previous two examples.

Now let’s check the contents of the table.

SELECT * FROM JsonCats3;

Result:

+-------+------------------------------------------------------+--------+
| key   | value                                                | type   |
|-------+------------------------------------------------------+--------|
| 0     | { "id" : 1, "name" : "Fluffy", "sex" : "Female" }    | 5      |
| 1     | { "id" : 2, "name" : "Long Tail", "sex" : "Female" } | 5      |
| 2     | { "id" : 3, "name" : "Scratch", "sex" : "Male" }     | 5      |
+-------+------------------------------------------------------+--------+

This table contains the three columns as mentioned. The value column contains each array element.

Example 4 – Use JSON Keys as Column Headers

This example is a bit of a cross between the previous two examples.

DECLARE @json NVARCHAR(4000) = N'{ 
    "pets" : {
            "cats" : [
            { "id" : 1, "name" : "Fluffy", "sex" : "Female" },
            { "id" : 2, "name" : "Long Tail", "sex" : "Female" },
            { "id" : 3, "name" : "Scratch", "sex" : "Male" }
        ],
            "dogs" : [
            { "name" : "Fetch", "sex" : "Male" },
            { "name" : "Fluffy", "sex" : "Male" },
            { "name" : "Wag", "sex" : "Female" }
        ]
    }
}';

SELECT * INTO JsonCats4
FROM OPENJSON(@json, '$.pets.cats')
WITH (
    id int,
    name varchar(60),
    sex varchar(6)
);

We are still defining our own schema, because we’re using the WITH clause. But you’ll notice that I’m not mapping the column names to any JSON path. This is because I’m using the actual names of the JSON keys.

When you do that, OPENJSON() is smart enough to match your column names to the JSON keys.

Let’s see what’s in the table.

SELECT * FROM JsonCats4;

Result:

+------+-----------+--------+
| id   | name      | sex    |
|------+-----------+--------|
| 1    | Fluffy    | Female |
| 2    | Long Tail | Female |
| 3    | Scratch   | Male   |
+------+-----------+--------+

So the data has been inserted into the table, just like in the first two examples, but this time the column names have been taken from the JSON document.

Example 5 – Specify Less Columns

You don’t need to include all values from the JSON document if you don’t need them all. You can specify just those that you need.

You can do this by specifying the columns in the SELECT list.

DECLARE @json NVARCHAR(4000) = N'{ 
    "pets" : {
            "cats" : [
            { "id" : 1, "name" : "Fluffy", "sex" : "Female" },
            { "id" : 2, "name" : "Long Tail", "sex" : "Female" },
            { "id" : 3, "name" : "Scratch", "sex" : "Male" }
        ],
            "dogs" : [
            { "name" : "Fetch", "sex" : "Male" },
            { "name" : "Fluffy", "sex" : "Male" },
            { "name" : "Wag", "sex" : "Female" }
        ]
    }
}';

SELECT 
    id, 
    name 
INTO JsonCats5a
FROM OPENJSON(@json, '$.pets.cats')
WITH (
    id int,
    name varchar(60),
    sex varchar(6)
);

SELECT * FROM JsonCats5a;

Result:

+------+-----------+
| id   | name      |
|------+-----------|
| 1    | Fluffy    |
| 2    | Long Tail |
| 3    | Scratch   |
+------+-----------+

Another way to do it is to remove the relevant column/s from the WITH clause.

DECLARE @json NVARCHAR(4000) = N'{ 
    "pets" : {
            "cats" : [
            { "id" : 1, "name" : "Fluffy", "sex" : "Female" },
            { "id" : 2, "name" : "Long Tail", "sex" : "Female" },
            { "id" : 3, "name" : "Scratch", "sex" : "Male" }
        ],
            "dogs" : [
            { "name" : "Fetch", "sex" : "Male" },
            { "name" : "Fluffy", "sex" : "Male" },
            { "name" : "Wag", "sex" : "Female" }
        ]
    }
}';

SELECT * INTO JsonCats5b
FROM OPENJSON(@json, '$.pets.cats')
WITH (
    id int,
    name varchar(60)
);

SELECT * FROM JsonCats5b;

Result:

+------+-----------+
| id   | name      |
|------+-----------|
| 1    | Fluffy    |
| 2    | Long Tail |
| 3    | Scratch   |
+------+-----------+

Although, it’s probably better to do both.

DECLARE @json NVARCHAR(4000) = N'{ 
    "pets" : {
            "cats" : [
            { "id" : 1, "name" : "Fluffy", "sex" : "Female" },
            { "id" : 2, "name" : "Long Tail", "sex" : "Female" },
            { "id" : 3, "name" : "Scratch", "sex" : "Male" }
        ],
            "dogs" : [
            { "name" : "Fetch", "sex" : "Male" },
            { "name" : "Fluffy", "sex" : "Male" },
            { "name" : "Wag", "sex" : "Female" }
        ]
    }
}';

SELECT 
    id,
    name
INTO JsonCats5c
FROM OPENJSON(@json, '$.pets.cats')
WITH (
    id int,
    name varchar(60)
);

SELECT * FROM JsonCats5c;

Result:

+------+-----------+
| id   | name      |
|------+-----------|
| 1    | Fluffy    |
| 2    | Long Tail |
| 3    | Scratch   |
+------+-----------+

Example 6 – Specify Less Rows

You can also use normal T-SQL syntax to filter the rows, so that only some records are inserted into the table. 

DECLARE @json NVARCHAR(4000) = N'{ 
    "pets" : {
            "cats" : [
            { "id" : 1, "name" : "Fluffy", "sex" : "Female" },
            { "id" : 2, "name" : "Long Tail", "sex" : "Female" },
            { "id" : 3, "name" : "Scratch", "sex" : "Male" }
        ],
            "dogs" : [
            { "name" : "Fetch", "sex" : "Male" },
            { "name" : "Fluffy", "sex" : "Male" },
            { "name" : "Wag", "sex" : "Female" }
        ]
    }
}';

SELECT 
    id,
    name
INTO JsonCats6
FROM OPENJSON(@json, '$.pets.cats')
WITH (
    id int,
    name varchar(60)
)
WHERE id IN (1,2);

SELECT * FROM JsonCats6;

Result:

+------+-----------+
| id   | name      |
|------+-----------|
| 1    | Fluffy    |
| 2    | Long Tail |
+------+-----------+

In this case I used a WHERE clause to insert just the rows that I’m interested in.

Import JSON from a File

You can use OPENJSON() in conjunction with the OPENROWSET() function to import a JSON file into a table.

This enables you to upload the data from a JSON file on a local drive or network drive. This saves you from having to copy and paste the document’s contents into your SQL code. This can be particularly beneficial when working with large JSON documents.