balerin 0.1.4

A python package startup orchestrator. it can handle loading all packages of an application at startup time respecting package dependencies.

Balerin

Your application will dance with Balerin!

Balerin is a python package startup orchestrator. it can handle loading all packages of an application at startup time respecting package dependencies.

Loading all application packages at startup has many benefits:

• Revealing all syntax errors and invalid imports on startup, preventing runtime failure.
• Initializing all required objects at startup, providing better performance at runtime.
• Triggering all first level objects of a module (ex. decorators) without the need to manually trigger them.
• Better code maintainability by letting you distribute the code on their suitable packages and preventing the need for a centralized location for nonsense imports (ex. models, hooks, managers, ...).
• Auto registration of celery tasks without the need to name all task modules a specific name (usually tasks).
• Populating application level caches at startup to boost performance at runtime.

Can I Use Balerin With?

Yes, you can. Balerin can be used in conjunction with all major frameworks (Flask, FastAPI, Sanic, AIOHTTP, Bottle, Pyramid, Tornado, Django) and many more. You can also use it on bare python applications without a framework, it's all up to you.

What about Pyrin?

It's a good question. Pyrin has builtin support for Balerin, so you can just use Pyrin without the need to prepare Balerin on your own.

Installing

Install using pip:

pip install balerin

Usage Example

There are two ways to use Balerin in your project:

• Basic: Loading all packages based on filesystem order. (the order could be changed between each run).
• Pro: Loading all packages respecting their dependencies on each other. (the order will be always the same on every run).

Sample Project Structure:

• root_dir
  • my_app
    • accounting
      • __init__.py
      • api.py
    • customers
      • __init__.py
      • models.py
    • __init__.py
    • api.py
    • models.py
  • start.py

Basic Usage:

my_app/__init__.py:

import os

from flask import Flask
from balerin import PackagingManager


app = Flask('my_app')
working_directory = os.path.abspath(os.getcwd())
root_package = os.path.join(working_directory, 'my_app')
balerin = PackagingManager(root_package, context=dict(important=True, app=app))

start.py:

from my_app import balerin, app

balerin.load_components()
app.run()

Pro Usage:

In order to load packages respecting their dependencies on each other, you must define a package class in __init__.py file of each package that has dependency on other packages. The package class must be a subclass of Balerin's Package class:

my_app/accounting/__init__.py:

from balerin import Package


class AccountingPackage(Package):
    # the name of the package.
    # example: `my_app.api`.
    # should get it from `__name__` for each package.
    NAME = __name__

    # list of all packages that this package is dependent
    # on them and should be loaded after all of them.
    # example: ['my_app.logging', 'my_app.api.public']
    # this can be left as an empty list if there is no dependencies.
    DEPENDS = ['my_app.customers']

    # specifies that this package is enabled and must be loaded.
    ENABLED = True

    # name of a module inside this package that should be loaded before all other modules.
    # for example: 'manager'
    # this can be left as None if there is no such file in this package needing early loading.
    COMPONENT_NAME = None

Now you can load your packages respecting their dependencies on each other, using the sample code in Basic Usage section.

How to Choose Between Basic or Pro Usages:

In most cases, you don't need to use the Pro Usage style. unless your application architecture is based on Dependency Injection and IoC concepts. so, when in doubt, go with Basic Usage.

Initialization Arguments of PackagingManager Class:

• root: it can be a single or multiple paths to different packages of your application. Balerin will load all sub-packages of each path respectively.
• base_component: specifies a module name which must be loaded before all other modules in each package if available. for example manager. this value could be overridden in each Package class using COMPONENT_NAME attribute.
• verbose: specifies that loading info should be printed on each step. defaults to True if not provided.
• ignored_packages: list of package names that must be ignored from loading. package names must be fully qualified. for example: my_app.accounting.public. notice that if a package that has sub-packages added to ignore list, all of its sub-packages will be ignored automatically even if not present in ignore list.
• ignored_modules: list of module names that must be ignored from loading. module names could be fully qualified or just the module name itself. for example: my_app.customers.models or models. notice that if only module name is provided, then all modules matching the provided name will be ignored from loading.
• ignored_detector: a function to be used to detect if a package or module should be ignored. it must take two arguments, the first is the fully qualified name and the second is a boolean value indicating that the input is a module. it also should take optional keyword arguments as context. it should return a boolean value. for example: my_detector(name, is_module, **context)
• module_loader: a function to be used to load custom attributes of a module. it should take two arguments, a name and a module instance.
  it also should take optional keyword arguments as context. the output will be ignored. for example: my_loader(name, module, **context)
• context: a dict containing all shared contexts to be used for example inside ignored_detector and module_loader functions.

PackagingManager Public Interface:

Once you create an object of PackagingManager class, you can call these methods on the created object:

• load_components: load all packages inside the root path.
• load: load a single module with provided name.
• get_loaded_packages: get a list of all loaded package names.
• is_package_loaded: get a value indicating that given package is loaded.
• get_context: get a dict of all shared contexts.

import os

from balerin import PackagingManager


working_directory = os.path.abspath(os.getcwd())
root_package = os.path.join(working_directory, 'my_app')
balerin = PackagingManager(root_package, context=dict(important=True))

balerin.load_components()
balerin.load('my_app.accounting.api')
loaded_packages = balerin.get_loaded_packages()
is_package_loaded = balerin.is_package_loaded('my_app.accounting')
context = balerin.get_context()

Hint

Balerin is an albanian word and means dancer.