json_tuple table-valued generator function
Applies to: 
Databricks SQL Databricks Runtime
Returns multiple JSON objects as a tuple.
Syntax
json_tuple(jsonStr, path1 [, ...] )
Arguments
jsonStr: ASTRINGexpression with well-formed JSON.pathN: ASTRINGliteral with a JSON path expression.
Returns
A single row composed of the JSON objects.
If any object cannot be found, NULL is returned for that object.
-
Applies to:
Databricks Runtime 12.1 and earlier:json_tuplecan only be placed in theSELECTlist as the root of an expression or following a LATERAL VIEW. When placing the function in theSELECTlist there must be no other generator function in the sameSELECTlist or UNSUPPORTED_GENERATOR.MULTI_GENERATOR is raised. -
Applies to:
Databricks SQL Databricks Runtime 12.2 LTS and above:Invocation from the LATERAL VIEW clause or the
SELECTlist is deprecated. Instead, invokejson_tupleas a table_reference.
Examples
Applies to: Databricks Runtime 12.1 and earlier:
> SELECT json_tuple('{"a":1, "b":2}', 'a', 'b'), 'Spark SQL';
1 2 Spark SQL
> SELECT json_tuple('{"a":1, "b":2}', 'a', 'c'), 'Spark SQL';
1 NULL Spark SQL
> SELECT json_tuple('{"a":1, "b":2}', 'a', 'c'), json_tuple('{"c":1, "d":2}', 'c', 'd'), 'Spark SQL';
Error: UNSUPPORTED_GENERATOR.MULTI_GENERATOR
Applies to: Databricks SQL Databricks Runtime 12.2 LTS and above:
> SELECT j.*, 'Spark SQL' FROM json_tuple('{"a":1, "b":2}', 'a', 'b') AS j;
1 2 Spark SQL
> SELECT j.*, 'Spark SQL' FROM json_tuple('{"a":1, "b":2}', 'a', 'c') AS j;
1 NULL Spark SQL
> SELECT j1.*, j2.*, 'Spark SQL'
FROM json_tuple('{"a":1, "b":2}', 'a', 'c') AS j1,
json_tuple('{"c":1, "d":2}', 'c', 'd') AS j2;