All Attributes from GeoJSON Retrieved via HTTPCaller (FME 2021)

There is no thing as dynamically exposing attributes in FME. When you need dynamic writing, the way to go is to use a Schema feature and feed this in the FeatureWriter. A Schema Feature is default output from the FeatureReader. You can also use the SchemaScanner to generate a Schema feature from datafeatures.

Can you update your FME ? In FME 2022, the SchemaScanner was released. It would greatly ease your dynamic workflow.

https://support.safe.com/hc/en-us/articles/25407488330509-Dynamic-Workflows-with-the-SchemaScanner

Else, you could try to set your HTTPCaller to download the result in files, and read them with a FeatureReader. The FeatureReader will output Schema features for each file you can use in a writer.


EDIT : if you’re not used to “dynamic workflow”, you should read about it there :
 https://support.safe.com/hc/en-us/articles/25407480725645-Tutorial-Dynamic-Workflows

The SchemaScanner is available in 2021. (At least in 2021.2.6)

Yes, this is probably the easiest way.

Attached working sample.

Hm, I’m pretty sure it was released in 2022. Here’s our article about it:
https://www.veremes.com/actus/sortie-de-fme-2022
(Sorry, it’s a French resource!)

It must have been backported then! Or we might have included this one in the 2022 new features if it was a late release of 2021.

It would be interesting to have the date and versions of transformers vs version of FME in the official documentation !

Thanks for the suggestions!

Unfortunately, upgrading is not an option at the moment since my customer is using FME 2021.0, and we don’t have access to the SchemaScanner.

I’ve already tried the workaround where I write the result to file and then use a FeatureReader to dynamically handle the schema. However, there are two major issues with this approach:

Many requests return no data, but users still expect a file to be generated for confirmation purposes. The FeatureReader throws hundreds of warnings when reading empty files, and this behavior is not acceptable to my customer.

There's a strange bug: when publishing a workspace that uses the FeatureReader, some other files on the server are randomly removed. We haven’t been able to fully identify the root cause, but this makes the approach unreliable in production.

Due to these limitations, the current workaround using the FeatureReader is not viable for us. We're looking into other ways to build a static schema or dummy feature to maintain consistency, but it's far from ideal.

Let me know if anyone has other ideas that work in 2021.0 without relying on FeatureReader or SchemaScanner.

This is at least something which will technically work, but I agree it will take some work.