Can you also please address all the Sonar findings?

@dreamorosi I'm still a bit confused about extending the KinesisDataStreamRecord.

if (eventType === EventType.KinesisDataStreams) {
      const extendedSchemaParsing = parse(record, undefined, schema, true);
      if (extendedSchemaParsing.success)
        return extendedSchemaParsing.data as KinesisStreamRecord;
      if (schema['~standard'].vendor === SchemaType.Zod) {
        const { JSONStringified } = await import(
          '@aws-lambda-powertools/parser/helpers'
        );
        const { KinesisDataStreamRecord } = await import(
          '@aws-lambda-powertools/parser/schemas/kinesis'
        );
        const extendedSchema = KinesisDataStreamRecord.extend({
          // biome-ignore lint/suspicious/noExplicitAny: The vendor field in the schema is verified that the schema is a Zod schema
          data: // Decompress and decode the data to match against schema
        });
        return parse(record, undefined, extendedSchema);
      }
      console.warn(
        'The schema provided is not supported. Only Zod schemas are supported for extension.'
      );
      throw new Error('Unsupported schema type');
    }

To extend it, should I create another helper which does the decompressions and decoding?
Or should I just use Envelopes for the whole thing. But, using Envelopes would just return the internal payload and our record handler expects the whole Record. We could maybe update the Record with the parsed payload and return it.
What would you suggest?

I've been thinking about your question for a while and there's no straightforward way to do it with our current feature set.

We can't use envelopes because of what you mentioned, and doing the parsing in two parts while possible, is suboptimal for two reasons:

• we'd be iterating through the entire batch in order to parse each record and then we'd have to recompose the object
• in case of parsing errors the path of the failed field(s) would be messed up since we'd no longer parse a schema (i.e. there's an error at the 2nd item in the batch, the error path should be something like Records.1.kinesis.data.foo)

With this in mind, I think we'll have to extract the transform logic that we have here into its own helper (with a dedicated PR) and then use the helper in the Batch Processing utility when extending.

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud