README.md

Function's Service Account should have role logging.writer

context.function_name DOES contain function id

Thanks to https://github.com/mcode-cc/python-yandex-cloud-logging library

main.py

import yandexcloud
from pyclm.logging import Logger

def handler(event, context):
    log = Logger(
        sdk=yandexcloud.SDK(),
        log_group_id="e23l2...",
        resource_type="serverless.function",
        resource_id=context.function_name,
        elements=1,
        period=0
    )

    log.error(
        "message test",
        request_id=context.request_id,
        version_id=context.function_version
    )
    
    return {'statusCode': 200}

requirements.txt

python-yandex-cloud-logging==0.1.0
# only for using_module.py
git+https://github.com/vitkhab/yc_cloud_logging_handler@0.1.0#egg=yc_cloud_logging_handler==0.1.0

using_logging.py

import yandexcloud
from pyclm.logging import Logger
import logging

logger = logging.getLogger(__name__)

# Without nullifying root logger handlers log dubles would be produced: one entry to 
# Cloud Logging and one entry to Function logs (which also be shown in Cloud Logging)
root_logger = logging.getLogger()
root_logger.handlers = []

class YandexCloudLoggingHandler(logging.Handler):

    def __init__(self, log_group_id, resource_id, request_id, function_version):
        logging.Handler.__init__(self)

        self.request_id = request_id
        self.function_version = function_version

        self.logger = Logger(
            sdk=yandexcloud.SDK(),
            log_group_id=log_group_id,
            resource_type="serverless.function",
            resource_id=resource_id,
            elements=1,
            period=0
        )
    
        self.loggers = {
            "TRACE": self.logger.trace,
            "DEBUG": self.logger.debug,
            "INFO": self.logger.info,
            "WARN": self.logger.warn,
            "ERROR": self.logger.error,
            "FATAL": self.logger.fatal
        }

    def emit(self, record):
        msg = self.format(record)

        if record.levelname in self.loggers:
            self.loggers[record.levelname](
                msg,
                request_id=self.request_id,
                version_id=self.function_version
            )
        else:
            self.loggers["DEBUG"](
                msg,
                request_id=self.request_id,
                version_id=self.function_version
            )

def push_info():
    logger.info("test info message")

def push_error():
    logger.error("test error message")

def handler(event, context):
    ycl_handler = YandexCloudLoggingHandler(
        log_group_id="e23l2k...",
        resource_id=context.function_name,
        request_id=context.request_id,
        function_version=context.function_version
    )

    formatter = logging.Formatter("%(filename)s on %(lineno)s: %(message)s")
    ycl_handler.setFormatter(formatter)
    
    # I presume Cloud function stays in memory for sometime. If there are several
    # requests in a row there will be several handlers added. So check is needed.
    for handler in logger.handlers:
        if isinstance(handler, YandexCloudLoggingHandler):
            break
    else:
        logger.addHandler(ycl_handler)
    logger.setLevel(logging.INFO)
    
    push_info()
    push_error()

    return {'statusCode': 200}

using_module.py

from yc_cloud_logging_handler import YandexCloudLoggingHandler
import logging

logger = logging.getLogger(__name__)

# Without nullifying root logger handlers log dubles would be produced: one entry to 
# Cloud Logging and one entry to Function logs (which also be shown in Cloud Logging)
root_logger = logging.getLogger()
root_logger.handlers = []

def push_info():
    logger.info("test info message")

def push_error():
    logger.error("test error message")

def handler(event, context):
    ycl_handler = YandexCloudLoggingHandler(
        log_group_id="e23l2...",
        resource_id=context.function_name,
        request_id=context.request_id,
        function_version=context.function_version
    )

    formatter = logging.Formatter("%(filename)s on %(lineno)s: %(message)s")
    ycl_handler.setFormatter(formatter)
    
    # I presume Cloud function stays in memory for sometime. If there are several
    # requests in a row there will be several handlers added. So check is needed.
    for handler in logger.handlers:
        if isinstance(handler, YandexCloudLoggingHandler):
            break
    else:
        logger.addHandler(ycl_handler)

    logger.setLevel(logging.INFO)
    
    push_info()
    push_error()

    return {'statusCode': 200}