1

I've got a dataframe that basically has a "key" column and a column that is a JSON string.

key jsonString
111 {"id" : "12345", "foo" : "stuff"}
111 {"id" : "23456", "bar" : "other stuff"}
111 {"id" : "34567", "baz" : "even other stuff"}

For each "key" value, I want to combine all the JSON strings into a JSON array, with a couple of other elements (with the ultimate goal of publishing to a Kafka topic). So I would end up with this:

{
    "type" : "combined",
    "values" :
    [
        {"id" : "12345", "foo" : "stuff"},
        {"id" : "23456", "bar" : "other stuff"},
        {"id" : "34567", "baz" : "even other stuff"}
    ]
}
    

I tried concatenating it all into a big string, but that went poorly. Is there a way to do this that doesn't involve having one giant schema for all my JSON strings? The real JSON is obviously much more complicated, and there are 4 possible schemas each one could be.

1 Answer 1

3

One way to get the JSON strings into an array is to group by key and use the collect_list function (pyspark, scala spark). The collect_set function could also be used if you wanted to eliminate duplicates.

df.show(false)
val groupedJson = df.groupBy("key").agg(collect_list("jsonString").as("jsonArray"))
groupedJson.show(false)
key jsonString
1 {"id": 123, "foo": "stuff"}
1 {"id": 456, "foo1": "stuff2"}
1 {"id": 789, "foo2": "stuff3"}
2 {"id": 000, "foo3": "stuff4"}
key jsonArray
1 [{"id": 123, "foo": "stuff"}, {"id": 456, "foo1": "stuff2"}, {"id": 789, "foo2": "stuff3"}]
2 [{"id": 000, "foo3": "stuff4"}]

I'm not sure if you wanted to put the array back into a JSON, but this would at least set you up to do that if needed. Here's another post that shows how you might do that: Create JSON column in Spark Scala

Sign up to request clarification or add additional context in comments.

1 Comment

Fantastic, this is super helpful, thanks

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.