Merging JSON outputs of parallel states in Step Function - amazon-web-services

Question

I have a parallel block in a Step Function. The input to the parallel block looks like
{ "a": null, "b": null }
The 2 parallel states are populateA and populateB. The output from populateA is
{ "a": "A", "b": null }
And the output from populateB is
{ "a": null, "b": "B" }
The output from the parallel block is an array
[ { "a": "A", "b": null }, { "a": null, "b": "B" } ]
How do I merge the two into
{ "a": "A", "b": "B" }

Answer 1

If PopulateA always only populates the value of a and PopulateB only the value of b, you can use a Pass state with Parameters (instead of the Result field) after your Parallel State to combine the inputs. Parameters allow you to transform your input by specifying key-value pairs using JSONPath.
Here is an example State Machine:
{
"StartAt": "Parallel",
"States": {
"Parallel": {
"Type": "Parallel",
"ResultPath": "$.CombinedOutput",
"Next": "MergeOutputs",
"Branches": [{
"StartAt": "populateA",
"States": {
"populateA": {
"Type": "Pass",
"Result": {
"a": "A",
"b": null
},
"End": true
}
}
},
{
"StartAt": "populateB",
"States": {
"populateB": {
"Type": "Pass",
"Result": {
"a": null,
"b": "B"
},
"End": true
}
}
}
]
},
"MergeOutputs": {
"Type": "Pass",
"Parameters": {
"a.$": "$.CombinedOutput[0].a",
"b.$": "$.CombinedOutput[1].b"
},
"Next": "EndState"
},
"EndState": {
"Type": "Pass",
"End": true
}
}
}

Answer 2

I'm afraid that it is not possible (currently) to do that without additional state. You can add additional Task after Parallel state which will merge array elements.
The code of the additional task would look like this (Python):
def merge_dicts(a, b):
result = a.copy()
for key, value in b.items():
if value is not None:
result[key] = value
return result
def handler(event, context):
result = {}
for item in event:
result = merge_dicts(result, item)
return result
handler takes output from the Parallel states and merges it into single dictionary. The state should be directly after Parallel.
Keep in mind that this code handles only flat-dict (not nested ones) so it will work in simple cases but not in complex ones (I've added it here only to show the general idea).