How to store logs in files depending on the execution group?

I have a number of functions that form a data processing pipeline. Data extraction, transformation, loading into the database.

In addition to abstraction by functional meaning, I want to separate the program logs depending on the execution group. For example, I have a pipelines: “store 1 sales”, “store 1 warehouses”, “store 2 sales”. The logic is something like this:

store1_extract_sales(): pass
store1_convert_sales(): pass
store2_load_sales(): pass

# same for other pipelines.

However, I would like all the logs to be divided into files “store1_sales”, “store1_warehouses” and “store2_sales”, and have their own rotation logic.

Moreover, data processing script files are located in different directories and files. (Extracting files) -> (Transforming files) -> Load file(Database worker)

I use the loguru library and now I just write all the logs into one file:

logger.add(
    sink=logs_directory + '/logs.json',
    level='INFO',
    rotation='30 day',
    serialize=True,
    encoding='utf-8'
)

How can i do this?

I tried to change the logger configuration and install a filter, but it didn’t work.

Solution

From the loguru documentation, you can add multiple log handlers, each to it's own file; and you create a filter for which logger to use. To connect a log handler to particular logger, you use the bind method.

# file 1
from loguru import logger

logger.add(
    "store1_sales.log", 
    filter=lambda rec: rec["extra"]["task"] == "store1-sales", 
    rotation="1 week"
)

logger_1 = logger.bind(task="store1-sales")

def store1_load_sales():
    logger_1.info('loading sales')
    ...

def store1_convert_sales():
    logger_1.info('converting sales')
    ...

Similarly for other files:

# file 2
from loguru import logger

logger.add(
    "store2_sales.log", 
    filter=lambda rec: rec["extra"]["task"] == "store2-sales", 
    rotation="1 week"
)

logger_2 = logger.bind(task="store2-sales")

def store2_load_sales():
    logger_2.info('loading sales')
    ...

def store1_convert_sales():
    logger_2.info('converting sales')
    ...