How to log output something like a structured dict object instead of a serialized string

[mirumirumi]

I am currently working on Rust on Lambda.
I am used to developing in Python & PowerTools and would like to produce similar structured logs in Rust.

However, when I started learning about loggers, I learned that normal logging is usually done with values in parallel directly under the fields key, and it is not common to log deeply nested objects as they are.

I have been logging dictionaries whose contents I want to look at as shown below, which is actually convenient (in this case, the logs appear in the message key, not in the fields key, but this is not important to me).

something_deep_dict = {
  ...
  {
     ...
  }
}

logger.info(something_deep_dict)

How does PowerTools output structured logs?
Do you have any special implementations?

Thank you!

[heitorlessa]

hey @mirumirumi, for Rust, I'd defer to @nmoutschen. For Python, we're not doing any special on that regard other than serializing to str any non-serializable value. Also, note that you can have additional fields at the root (in parallel to message key), when you pass additional key-value pairs after the logging message.

Here's a quick example I ran locally:

from aws_lambda_powertools import Logger
from typing import Any

logger = Logger(service="discussion-2086")

nested_dict: dict[str, Any] = {
    "products": [
        {"id": "8dbc54d1-6e5c-4f9b-8f4c-2a5fd3e2b2a7", "description": "dummy"},
    ],
    "customer_id": "95126722-4665-4a44-8150-8e12233c9bcd",
}

logger.info("Incoming products", some_key="some_value", **nested_dict)

JSON output

{
    "level": "INFO",
    "location": "<module>:13",
    "message": "Incoming products",
    "timestamp": "2023-04-05 17:14:46,838+0200",
    "service": "discussion-2086",
    "some_key": "some_value",
    "products": [
        {
            "id": "8dbc54d1-6e5c-4f9b-8f4c-2a5fd3e2b2a7",
            "description": "dummy"
        }
    ],
    "customer_id": "95126722-4665-4a44-8150-8e12233c9bcd"
}