Only replace the logging formatter if it is a logging.Formatter (DataDog#155)

* Only replace the logging handler if the logging handler is using a logging.Formatter

* Update README to mention that DD_LOGS_INJECTION only replaces the logger.Formatter in the LambdaLoggerHandler

Author: agocs

README.md

@@ -65,7 +65,7 @@ If `DD_FLUSH_TO_LOG` is set to `false` (not recommended), you must set `DD_SITE`
 
 ### DD_LOGS_INJECTION
 
-Inject Datadog trace id into logs for [correlation](https://docs.datadoghq.com/tracing/connect_logs_and_traces/python/). Defaults to `true`.
+Inject Datadog trace id into logs for [correlation](https://docs.datadoghq.com/tracing/connect_logs_and_traces/python/) if you are using a `logging.Formatter` in the default `LambdaLoggerHandler` by the Lambda runtime. Defaults to `true`.
 
 ### DD_LOG_LEVEL
 

datadog_lambda/tracing.py

@@ -336,7 +336,10 @@ def inject_correlation_ids():
     # Override the log format of the AWS provided LambdaLoggerHandler
     root_logger = logging.getLogger()
     for handler in root_logger.handlers:
-        if handler.__class__.__name__ == "LambdaLoggerHandler":
+        if (
+            handler.__class__.__name__ == "LambdaLoggerHandler"
+            and type(handler.formatter) == logging.Formatter
+        ):
             handler.setFormatter(
                 logging.Formatter(
                     "[%(levelname)s]\t%(asctime)s.%(msecs)dZ\t%(aws_request_id)s\t"