My array has data but is coming out with a type of null. Can someone explain why and how to fix?

[gyjhfyjdrfhj]

Hoping someone can point me in the right direction. My flow is doing the following:

• Filter Array: Filter my_array for item()?['Errors'] is not equal to None (the field says "None" if there are no errors)  (apparently this is the problem)
• Set variable my_errors to Body of Filter Array -- seems to works FINE
• Filter Array #2: Filter my_array for item()?['Errors'] is equal to None -- works FINE
• Set variable my_array to Body of Filter Array #2 -- works FINE
• Parse JSON of my_array, create CSV and send it somewhere -- works FINE
• If length of error_array> 0, Parse JSON of my_errors -- FAIL because the schema is wrong... I go check how that is possible and I see this:

For some reason instead of "body" is says null when the array gets filtered. The array is not null and has all the records that it should have. The weird thing is that the second filter condition is doing the exact same thing but is resulting in "body" instead of null, so it gets parse properly. 

{"body":[{"Fiscal Year":2022....

[Koen5]

Dear gyjhfyjdrfhj,

It seems you have already pinpointed the issue ... the first filter seems to do stuff we do not expect.

My guess is that the "Filter my_array for item()?['Errors'] is not equal to None" is a syntax which means something else to PowerAutomate ... maybe because 'None' is something predefined?

I just ran the same issue through ChatGPT and it gives me the output to use None in the filter in like this: "None".

As such you define that JSON treads None as a value.

So can you try with the filter array as such: [something == "None" ]

Hope it helps,

Happy flowing,

Koen

[gyjhfyjdrfhj · Accepted solution]

I changed "None" to "" for my Error column, but the problem still happened. But I dug into the issue a bit more. Initially I am getting data from an HTTP call and storing that data in my_array. Each HTTP call returns an array, so I start out with an array of arrays. I need to unpack this into an array of records, and the fastest way to do that is to do a Compose with the following: json(concat('[', replace(replace(join(variables('new_array'), ','), '[', ''), ']', ''),']'))

And store the output as my_array. The problem is that clearly a null array made its way into my_array at some point, which is how I am ending up with the null.

Now, I either need to handle the possibility of a null array earlier in the code or I can modify my filter to filter out the nulls if they occur. I opted for the latter by modifying my filter to the following: and(not(equals(item()?['Errors'], null)), not(equals(item()?['Errors'], '')))