JSON is a human-readable, text-based format for exchanging semistructured data. Its hierarchical key–value structures can be loaded into Snowflake as VARIANT values without a predefined relational schema. section 4.6
A Snowflake pipeline can process JSON by:
- Ingesting JSON from object storage into a VARIANT column, subject to a 16 MB size limitation. section 4.6
- Selecting nested values with the colon operator, then converting returned VARIANT values to required data types. section 4.6
- Applying FLATTEN with LATERAL to turn repeated objects, such as order lists, into relational rows. section 4.6
Snowpark connection configuration
A Python program can read account, user, password, role, warehouse, database, and schema settings from a JSON file with json.load(), pass the resulting dictionary to Session.builder.configs(...).create(), and close the session afterward. Keep the credentials file outside the repository or exclude it with .gitignore. section 6.3.3
{
"account" : "ab12345",
"user" : "my_user",
"password" : "my_pass",
"role" : "SYSADMIN",
"warehouse" : "BAKERY_WH",
"database" : "BAKERY_DB",
"schema" : "SNOWPARK"
}
Use the configuration listing to inspect the JSON object structure and the connection parameters used by a Snowpark session. section 6.3.3
