Process complex json array member having null with Dataflow

Question

Below is the json data that is received via API call.

[
{
"resultsMetadata": {
"count" :100
"entity" : "test"
},
"results": [
{
"uniqueid": "100ABC",
"fnbr": 1,
"relations" : [
{
"relcode": "A1"
"reldesc": "A1 test"
},
{
"relcode": "A2"
"reldesc": null
},
{
"relcode": "A3"
"reldesc": "A3 test"
}
] # end of relations
},
{
"uniqueid": "101ABC",
"fnbr": 1,
"relations" : [
{
"relcode": "A1"
"reldesc": "A1 test"
},
{
"relcode": "A3"
"reldesc": null
},
{
"relcode": "A4"
"reldesc": "A4 test"
}
] # end of relations
}
] #end of results
}
]

Aim is to flatten this json response and store as parquet file.
Having tried with dataflow (source -> flatten -> sink), data gets flatten and new columns like relcode, reldesc are generated as expected.
But the data inside these new fields for first record "uniqueid": "100ABC" is stored as
relcode reldesc
[1] A1 [1] A1test
[2] A2 [2] A3test
[3] A3
This mean, array members that have null are not stored after flatten.
If there a way where I check this array member value and replace them with empty string ("").
Also, I expect the schema changes and so would like to have a generic and dynamic solution thereby it can check this for each array member that has null.