Python lambda 多行日志格式 output 到 cloudwatch

Question

I can override the default Lambda python log format like so:

LOG_FORMAT = '[%(levelname)s] %(asctime)s.%(msecs)dZ [%(filename)s] [%(funcName)s] %(message)s'
DATETIME_FORMAT = '%Y-%m-%dT%H:%M:%S'
logging.basicConfig(format=LOG_FORMAT, level=logging.INFO, datefmt=DATETIME_FORMAT, force=True)
LOG = logging.getLogger()

But I've noticed that the log lines will print onto separate lines in cloudwatch based on newline characters.

If I try and log a formatted XML document (to Cloudwatch) I end up, effectively, with new log lines for every line in the XML document.

But prior to modifying the default format the XML document would appear as a single log line, that I could easily copy & paste out of Cloudwatch

Prior to format change:

[INFO] 2022-11-30T02:16:54.345Z dc2518d8-60f3-461c-812b-c70b1b836592
SNS message: <?xml version="1.0"?>
<Document>
  ...
</Document>

In cloudwatch clicking on the default output displays the entire document, where it can be easily copied.

After format change:

[INFO] 2022-11-29T11:40:47.563Z [function.py] [handler] SNS message: 
2022-11-29T11:40:47.563Z <?xml version="1.0"?>
2022-11-29T11:40:47.563Z <Document>
2022-11-29T11:40:47.563Z  ...
2022-11-29T11:40:47.563Z</Document>

Is the default formatter parsing the string and replacing newline characters? I know \r is treated differently to \n in CloudWatch.

This is the Lamba Python runtime bootstrap code:

https://github.com/aws/aws-lambda-python-runtime-interface-client/blob/main/awslambdaric/bootstrap.py

Answer 1

Figured out a workaround by editing the existing formatter:

LOG = logging.getLogger()
LOG.setLevel(logging.INFO)
log_handler = LOG.handlers[0]
log_handler.setFormatter(logging.Formatter('[%(levelname)s] %(asctime)s.%(msecs)dZ [%(filename)s] [%(funcName)s] %(message)s\n'))