Loading JSON data into Snowflake

Reading Time: 4 minutes

Have you ever faced any use case or scenario where you’ve to load JSON data into the Snowflake? We better know JSON data is one of the common data format to store and exchange information between systems. JSON is a relatively concise format. If we are implementing a database solution, it is very common that we will come across a system that provides data in JSON format. Snowflake has a very straight forward approach to load JSON data. In this blog, we will understand this approach in a step-wise manner.

1. Stage the JSON data

In snowflake Staging the data means, make the data available in Snowflake stage(intermediate storage) it can be internal or external. Staging JSON data in Snowflake is similar to staging any other files. Let’s Staging JSON data file from a local file system.

CREATE OR REPLACE STAGE my_json_stage file_format = (type = json);
PUT file:///home/knoldus/Desktop/family.json @my_json_stage;

We can also create an external stage using the AWS S3 bucket, or Microsoft Azure blob storage that contains JSON data.

CREATE OR REPLACE STAGE my_json_stage url='url of s3 bucket or azure blob with credentials'

The JSON data looks like:

  "Name": "Aman Gupta",
  "family_detail": [
      "Name": "Avinash Gupta",
      "Relationship": "Father",
      "Name": "Lata Gupta",
      "Relationship": "Mother",
      "Name": "Shrishti Gupta",
      "Relationship": "Sister",
      "Name": "Bobin Gupta",
      "Relationship": "Brother",

2. Load JSON data as raw into temporary table

To load the JSON data as raw, first, create a table with a column of VARIANT type. VARIANT can contain any type of data so it is suitable for loading JSON data.

CREATE TABLE relations_json_raw (
  json_data_raw VARIANT

Now let’s copy the JSON file into relations_json_raw table.

COPY INTO relations_json_raw from @my_json_stage;

Note that a file format does not need to be specified because it is included in the stage definition.

3. Analyze and prepare raw JSON data

The next step would be to analyze the loaded raw JSON data. Determining what information needs to be extracted from JSON data. For example, in our case, we are interested to extract the name key and from the family_detail array object, we want to extract the name and relationship key from each JSON object. The below query will do that.

    , lateral flatten( input => json_data_raw:family_detail );

The above query using lateral join and a flatten function. The flatten function returns a row for each JSON object from the family_detail array. and the lateral modifier joins the data with any information outside of the object, in our example candidate name that we are extracting with json_data_raw: Name.

Load Data into target table

Now we have analyzed and extracted information. We can load the extracted data into the target table.

CREATE OR REPLACE TABLE candidate_family_detail AS
    json_data_raw:Name AS candidate_name,
    VALUE:Name::String AS relation_name,
    VALUE:Relationship::String AS relationship

    , lateral flatten( input => json_data_raw:family_detail );

If you don’t want to do a “create table as”, you can pre-create a table and then insert the JSON data into the table.

Note: If you are loading JSON data recursively, the process needs to be setup in such a way that you can identify which row’s already exists in the target table or which row’s are new.

Thanks for reading. Stay connected for more future blogs.

Written by 

Exploring Big Data Technologies.