Unit testing in Python can be a bit tricky, especially when you’re dealing with multiple layers of nested function calls. Since you mentioned that you want to create mocks for a service that interacts with a database and has several components, here are some tips that might help you out.

One way to simplify mocking nested return values is to use unittest.mock.Mock along with return values that you can define for each level. Instead of dealing with side_effect, which can get messy, try to structure your mocks with return_value that directly matches what you’re expecting at each layer.

from unittest.mock import Mock

# Mocking the repository that fetches user data
user_repo_mock = Mock()
user_repo_mock.fetch_user.return_value = {'id': 1, 'name': 'John Doe'}

# Mocking the service that fetches user profile details
user_profile_service_mock = Mock()
user_profile_service_mock.get_profile.return_value = {
    'age': 30,
    'location': 'New York'
}

# Now, you can create the top-level service that uses both mocks
report_service = Mock()
report_service.generate_report.return_value = (
    user_repo_mock.fetch_user(), 
    user_profile_service_mock.get_profile()
)

# When you call the report_service, it returns expected values from mocks
mocked_user = report_service.generate_report()
print(mocked_user)  # This should print out the combined results
    

This way, each mock knows what to return without needing numerous levels of side_effect. Just make sure that the structure of your return values lines up with what your actual code expects.

Another pattern you might like is to use dictionaries or data classes to represent complex return types. This way, you can set up a clean, nested structure without cluttering your mocks with complicated logic.

from unittest.mock import Mock

# Sample data structure
class User:
    def __init__(self, id, name, profile):
        self.id = id
        self.name = name
        self.profile = profile

class Profile:
    def __init__(self, age, location):
        self.age = age
        self.location = location

# Setting up mocks with data classes
user_repo_mock = Mock()
user_repo_mock.fetch_user.return_value = User(1, 'John Doe', Profile(30, 'New York'))

# Call your function which now returns a structured object
user = user_repo_mock.fetch_user()
print(user.name, user.profile.age)  # Outputs: John Doe, 30
    

By using these methods, you can avoid the messiness of side effects and keep your tests clear and maintainable. Good luck with your unit testing journey!