[receiver/awsfirehose] Implement encoding extension framework

[axw]

Component(s)

receiver/awsfirehose

Is your feature request related to a problem? Please describe.

awsfirehosereceiver has a limited set of baked-in unmarshallers: cwlogs (CloudWatch logs), cwmetrics (CloudWatch metric streams JSON format), otlp_v1 (CloudWatch metric streams OTLP 1.0 format).

Firehose itself does not know or care about the format of data records. In theory you could write OTLP logs to a delivery stream using https://docs.aws.amazon.com/firehose/latest/APIReference/API_PutRecord.html, but there is no way for the receiver to unmarshal this. OTLP is just an example: one might produce arbitrary observability data (or other types of data, but out of scope here) to Firehose, which also cannot be decoded without additional unmarshallers.

The unmarshallers within the awsfirehosereceiver module may also be useful for other receivers, e.g. in the S3 receiver for decoding data sent by Firehose to an S3 bucket. Since they are internal to the module, they cannot be reused.

Describe the solution you'd like

I would like to enhance the receiver to implement the encoding extension framework, so that users may inject additional encodings. I have created an untested PoC of this at axw@1904aa5. In that PoC I have left the existing unmarshallers in the receiver, but I would envisage them eventually moving to dedicated extension components.

Describe alternatives you've considered

No response

Additional context

No response

[github-actions]

Pinging code owners:

• receiver/awsfirehose: @Aneurysm9

See Adding Labels via Comments if you do not have permissions to add labels yourself.

[axw]

If this direction is acceptable, I would be happy to polish and contribute the branch I referenced above.

[VihasMakwana]

This sounds like a valid request to me.

FWIW, there are many components that use custom unmarshalers/marshalers. Now that we have encoding extension, we should do this for all such components.