Logger clear_state flip removing structure keys

Using the clear_state argument in the logger decorator for lambda handlers is destroying the default structured keys.

What were you trying to accomplish? Just trying to get some lambda logs with custom keys, which ideally would be cleaned up after each warm call, such that those custom key values do not trickle to the next invocation.

Expected Behavior

Using the clear_state argument should not remove the default structured keys. Those should remain fixed. Only custom keys should be removed upon new invocation.

Current Behavior

Currently, default structured keys are being removed from the get-go.

Steps to Reproduce (for bugs)

First, this:

from aws_lambda_powertools import logging
from aws_lambda_powertools.utilities.typing import LambdaContext


logger = logging.Logger()

@logger.inject_lambda_context(clear_state=False)
def lambda_handler(event: dict, context: LambdaContext):
    logger.info("HEREEE")

Output:

{
    "level": "INFO",
    "location": "lambda_handler:21",
    "message": "HEREEE",
    "timestamp": "2022-03-29 17:53:56,985+0000",
    "service": "service_undefined",
    "cold_start": true,
    "function_name": "alex-test-layers",
    "function_memory_size": "1024",
    "function_arn": "arn:aws:lambda:us-west-2:111111111111:function:alex-test-layers",
    "function_request_id": "a1453466-2abe-4a55-ad9e-640e745f4133",
    "xray_trace_id": "1-624347b4-589b3226782e49797249edb1"
}

Now this clear_state=True

from aws_lambda_powertools import logging
from aws_lambda_powertools.utilities.typing import LambdaContext


logger = logging.Logger()

@logger.inject_lambda_context(clear_state=True)
def lambda_handler(event: dict, context: LambdaContext):
    logger.info("HEREEE")

Output:

{
    "message": "HEREEE",
    "service": "service_undefined",
    "cold_start": true,
    "function_name": "alex-test-layers",
    "function_memory_size": "1024",
    "function_arn": "arn:aws:lambda:us-west-2:111111111111:function:alex-test-layers",
    "function_request_id": "e3cf7717-88ea-43dc-b80c-57754bc6b8c9",
    "xray_trace_id": "1-62435f24-54e5827e68d838640494dca4"
}

Environment

• Powertools version used: aws-lambda-powertools==1.25.5
• Packaging format (Layers, PyPi): Layers
• AWS Lambda function runtime: Python 3.8
• Debugging logs: Above.

[eliasbrange]

@heitorlessa PR is up and running! #1095

[heitorlessa]

Thanks to your super high quality bug report we managed to reproduce and improved tests as part of the fix - the latest version 1.25.6 has this fix now. Lambda Layers upcoming version 16 will take a few hours to be available in multiple AWS commercial regions.