Skip to main content

Load environment variable based profiles in Python 3.6+ with ease

Project description

Introduction

wr-profiles allows declaring a profile – a collection of key-value pairs – backed by environment variables and accessing the profile contents through dot notation with auto-complete working in your IDE. You can also have multiple separate environments declared and switch between profiles by changing just one environment variable (just like you normally do when working with multiple AWS accounts if you have that background). wr-profiles are designed with testability in mind.

wr- in the package name stands for Wheel Reinvented. Just like all other packages whose name starts with wr-, the meaning of this library (as in meaning of life) lies in its existence and evolution and not in any external application of it.

Supported Python Versions

  • Python 3.6

  • Python 3.7

Example

Declare the profile:

from wr_profiles import envvar_profile_cls

@envvar_profile_cls
class WarehouseProfile:
    host: str = "localhost"
    username: str
    password: str

warehouse_profile = WarehouseProfile()

Set up the environments:

export WAREHOUSE_STAGING_PARENT_PROFILE="production"
export WAREHOUSE_STAGING_PASSWORD="staging-password"
export WAREHOUSE_PRODUCTION_USERNAME="production-username"
export WAREHOUSE_PRODUCTION_PASSWORD="production-password"

Select the active environment:

export WAREHOUSE_PROFILE=staging

Use the profile:

from profiles import warehouse_profile

assert warehouse_profile.host == "localhost"
assert warehouse_profile.username == "production-username"
assert warehouse_profile.password == "staging-password"

Installation

pip install wr-profiles

If you decide to use this library, make sure you pin the version number in your requirements file.

We are following interpretation of the semantic versioning schema. Example:

  • v2.x.a -> v2.x.b - bugfix or non-breaking change, safe to upgrade.

  • v2.x.* -> v3.y.* - potentially breaking changes, feature added, minimal changes to user code may be required

  • v2.* -> v3.* - complete changeover

Changelog

v4.2.0

  • Added profile_delegate feature.

v4.1.0

  • Added EnvvarProfile.create_env which creates an Environment which can be applied as a context manager.

User Guide

Concepts

Profile

A profile represents a set of configurable properties of a single service backed by environment variables.

In your application, there can be multiple unrelated profiles each providing interface to properties of a different service.

Instances of profiles associated with the same service share the same base class and are identified by profile_root specified in that base class. Is is the root from which all relevant environment variable names are formed.

Profiles of unrelated services do not share any information. In the discussion below, different instances or kinds of profiles all relate to the same service, e.g. same profile_root.

Warehouse Profile (Example)

In the discussion below, we will use a profile for a data warehouse access as an example. Class WarehouseProfile declares the profile and the properties it provides. Object warehouse_profile is the single instance through which user must look up service’s active configuration.

from wr_profiles import envvar_profile_cls

@envvar_profile_cls
class WarehouseProfile:
    host: str = "localhost"
    username: str
    password: str

warehouse_profile = WarehouseProfile()

Profile Name

Individual instances of profiles are identified by their name (profile_name property).

Active Profile

The active profile is the profile of a service that should be used according to the environment variables.

By default, the active profile can be switched by setting a special environment variable <PROFILE_ROOT>_PROFILE. For WarehouseProfile that would be WAREHOUSE_PROFILE.

The name of this variable can be customised by setting your class’s profile_activating_envvar.

If this variable is not set, the active profile is an empty string, and the environment variables consulted are in the form:

<PROFILE_ROOT>_<PROPERTY_NAME>

For example, WAREHOUSE_HOST.

If <PROFILE_ROOT>_PROFILE is set then the active profile consults environment variables in the form:

<PROFILE_ROOT>_<PROFILE_NAME>_<PROPERTY_NAME>

For example, if WAREHOUSE_PROFILE is set to staging then host property will be looked up under WAREHOUSE_STAGING_HOST.

Parent Profile

Any particular profile (for example, staging profile of WarehouseProfile) can be instructed to inherit its property values from a parent profile by setting:

<PROFILE_ROOT>_<PROFILE_NAME>_PARENT_PROFILE

For example, WAREHOUSE_STAGING_PARENT_PROFILE, if set to production, would mean that if environment variable WAREHOUSE_STAGING_HOST was not set, property value loader would consult WAREHOUSE_PRODUCTION_HOST instead. And only if that variable was not present, the default value of the property would be used.

Limitation: The default profile (profile_name="") cannot be used as a parent profile. If you specify empty string as <PROFILE_ROOT>_<PROFILE_NAME>_PARENT_PROFILE then this profile won”t have any parent profile. It is the same as having no value set.

Live Profile vs Frozen Profile

A live profile always consults environment variables (os.environ) whereas a frozen profile does so only during instantiation and when explicitly loaded with load() method.

Common Scenarios

Get Current Active Profile

Current active profile is always available through the instance of your profile class which is instantiated with no arguments:

warehouse_profile = WarehouseProfile()

Normally you’d only need a single instance of your profile class pointing to the active profile.

Get Concrete Profile

To work with a concrete profile which may not necessarily be activated, use load factory method:

staging = WarehouseProfile.load("staging")

By default, this profile will be frozen which means it will be loaded only once during instantiation. If you want it to always consult environment variables, pass profile_is_live=True:

staging = WarehouseProfile.load("staging", profile_is_live=True)

Customise Profile-Activating Environment Variable

Imagine you have your WarehouseProfile and you want to use it in tests. In tests it should have different defaults.

@envvar_profile_cls
class WarehouseTestProfile(WarehouseProfile):

    # If you don't set this, it would be "WAREHOUSE_PROFILE" which would conflict
    # with your non-test profile.
    profile_activating_envvar = "WAREHOUSE_TEST_PROFILE"

    host: str = "test-host"
    username: str = "test-user"

In your application you would then have two instances:

profile = WarehouseProfile()
test_profile = WarehouseTestProfile()

Now you can reuse your non-test profiles in tests when it makes sense. For example, if you have set up environment variables in the form WAREHOUSE_SANDBOX_* then this “sandbox” profile can be used in tests by just setting WAREHOUSE_TEST_PROFILE to sandbox.

Note that profile_root for both profiles is the same.

Activate Profile

To activate a profile, call activate method on a frozen instance of the profile without any arguments, or, activate(profile_name) on the live current profile instance:

staging.activate()
# or:
warehouse_profile.activate("staging")

Get All Values

warehouse_profile.to_dict()

Set Environment Variables

Note that the environment variables you set normally apply only to the current process and its sub-processes so this will have limited use – it will only make sense when you are launching sub-processes or you do this somewhere early in the code before environment variables are loaded by other parts of your code.

os.environ.update(warehouse_profile.to_envvars())

Check If Property Has Non-Default Value

warehouse_profile.has_prop_value("username")
# or
warehouse_profile.has_prop_value(WarehouseProfile.username)

Inspect Property

from wr_profiles import EnvvarProfileProperty

assert isinstance(WarehouseProfile.username, EnvvarProfileProperty)
assert WarehouseProfile.username.name == "username"
assert WarehouseProfile.username.default == "default-username"

Environment Objects

Starting from version 4.1 you can create an instance of Environment which can then be applied on os.environ or pytest’s monkeypatch fixture. Environment is a dictionary of environment variables that neet to be set or unset in order to apply the specified environment. The values are determined at environment creation time.

test_env = warehouse_profile.create_env(username='test', password=None)
with test_env.applied():
    assert warehouse_profile.username == 'test'
    assert os.environ['WAREHOUSE_USERNAME'] == 'test'

    assert warehouse_profile.password is None
    assert 'WAREHOUSE_PASSWORD' not in os.environ

Config Object that Delegates to Profile

Environment variables are not necessarily the only source of configuration. In a non-trivial application you probably won’t be consulting the profile object from your application code directly. Instead, you’ll have a config object which will consult different sources including the environment variable profile. It is very likely that the properties defined in your profile class will be a subset of those exposed by the config object. If you don’t want to repeat yourself, you can have your config class extend the profile class and have the config class delegate all the attributes to the profile class except for those implemented in the config class.

profile = WarehouseProfile()

class WarehouseConfig(WarehouseProfile):
    @property
    def profile_delegate(self):
        return profile

    @property
    def username(self):
        return profile.username or "anonymous"

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

wr-profiles-4.2.1.macosx-10.13-x86_64.tar.gz (15.5 kB view details)

Uploaded Source

Built Distribution

wr_profiles-4.2.1-py3-none-any.whl (9.7 kB view details)

Uploaded Python 3

File details

Details for the file wr-profiles-4.2.1.macosx-10.13-x86_64.tar.gz.

File metadata

  • Download URL: wr-profiles-4.2.1.macosx-10.13-x86_64.tar.gz
  • Upload date:
  • Size: 15.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/1.12.1 pkginfo/1.4.2 requests/2.20.0 setuptools/40.4.3 requests-toolbelt/0.8.0 tqdm/4.28.1 CPython/3.6.6

File hashes

Hashes for wr-profiles-4.2.1.macosx-10.13-x86_64.tar.gz
Algorithm Hash digest
SHA256 18be7ec6ac7aece69e65a98b429dc93406fe99b8fbcae19197c3afbc48257140
MD5 e8c8af76e183d6d94fe19335f53e2728
BLAKE2b-256 078190b6d64aab0279d528bd4688a77131f38b8a76692174a4bd211c455dcb89

See more details on using hashes here.

File details

Details for the file wr_profiles-4.2.1-py3-none-any.whl.

File metadata

  • Download URL: wr_profiles-4.2.1-py3-none-any.whl
  • Upload date:
  • Size: 9.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/1.12.1 pkginfo/1.4.2 requests/2.20.0 setuptools/40.4.3 requests-toolbelt/0.8.0 tqdm/4.28.1 CPython/3.6.6

File hashes

Hashes for wr_profiles-4.2.1-py3-none-any.whl
Algorithm Hash digest
SHA256 40167b05c757042e7693a647aa7de12dfe01715916961634c351c1a38f5a9074
MD5 1612076bc14e08627296c6f96cd55609
BLAKE2b-256 db97044a5e3ee3f49266b9768f26c34a8d951f30b5553b5a28e0b03e6af9c1f8

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page