Skip to main content

Yet another wrapper object for environment variables. Does only the things I care about.

Project description

Author:

Keryn Knight

Version:
0.2.2

A small module for wrapping over environment variables (pulled from os.environ) which provides convenience methods to fetch and check various data types (including iterables) in what I’d charitably hope is a sensible way.

Explicitly doesn’t attempt to read from any .env or .envrc file, because that doesn’t describe valid examples or which things may/should be set into the environment. It becomes an absolute pot-luck.

Tracks requested environment variables and their default/fallback/example values, and whether or not the fallback was used. Never tracks the actual environment value.

If this package isn’t to your liking, there’s plenty of others, and I’m largely suffering from Not-Invented-Here syndrome.

All methods exposed by the Environment accept a key and a default.

  • The key is the environment variable to search for.

  • The default MUST be a string, as it is subject to the same parsing as if it had been found in the environment, and thus serves as a documented example of a valid value to export as an environment variable. Enforced value documentation!

A series of examples

A short overview of all of the available check/cast methods on an Environment follows

Assume all examples are prefixed with:

from enviable import env

which is roughly equivalent to:

from enviable import Environment
import os
env = Environment(os.environ)

Remember, the second argument (default) is always a string, and always gets parsed the same as a real value, so treat it as an example value in the following…

Conversions and validations

To trim any accidental quotes or whitespace from the beginning and end of the value:

env.text("VAR_NAME", "'   test '") == "test"

To convert an incoming string to an integer:

env.int("VAR_NAME", "3") == 3

To convert an incoming string to real boolean (True or False), note that upper or lower case doesn’t matter:

env.bool("VAR_NAME", "true") is True
env.bool("VAR_NAME", "on") is True
env.bool("VAR_NAME", "1") is True
env.bool("VAR_NAME", "yes") is True
env.bool("VAR_NAME", "y") is True

env.bool("VAR_NAME", "false") is False
env.bool("VAR_NAME", "off") is False
env.bool("VAR_NAME", "0") is False
env.bool("VAR_NAME", "no") is False
env.bool("VAR_NAME", "n") is False

To make a uuid.UUID from an optionally hyphenated string:

env.uuid("VAR_NAME", "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa") == UUID('aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa')
env.uuid("VAR_NAME", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa") == UUID('aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa')

To check if an email looks valid:

env.email("VAR_NAME", "a@b.com") == "a@b.com"

To verify if a string is all hexi characters:

env.hex("VAR_NAME", "abcdef12345ABCDEF") == "abcdef12345ABCDEF"

Confirm that a string can be decoded from a base64 encoded value:

env.b64("VAR_NAME", "d29vZg==") == 'd29vZg=='

There’s no support for float because it’s lossy, but you can have decimals:

env.decimal("VAR_NAME", "1.25") == Decimal("1.25")

To confirm that a string looks like it might be an importable python thing:

env.importable("VAR_NAME", "path.to.my.module") == "path.to.my.module"

To make sure a string represents an existing, readable file on disk:

env.filepath("VAR_NAME", "/path/to/my/valid_file.json") == "/path/to/my/valid_file.json"

To make sure a string is a directory which exists:

env.directory("VAR_NAME", "/path/to/my") == "/path/to/my"

To vaguely sanity-check URLs (must start with http:// or https:// or // or /...):

env.web_address("VAR_NAME", "http://example.com/")

To constrain a value to one of a few valid options (where choices is parsed the same way as Iterables):

env.one_of("VAR_NAME", "3", choices="1,2,3,4")

and to go off-reservation, you can get JSON out, or the raw environment string:

env.json("VAR_NAME", "{}") == {}
env.raw("VAR_NAME", "'   ...  '") == "'   ...  '"

Temporal values (datetimes, dates, times)

If you have Django installed (because that’s my main use case and I’m lazy) you can also get datetimes if you provide a value in ISO 8601 format:

env.datetime("VAR_NAME", "2019-11-21 16:12:56.002344")
env.datetime("VAR_NAME", "2019-11-21 16:12:56.002344+20:00")
env.datetime("VAR_NAME", "2019-11-21")

Similarly you can ask for dates:

env.date("VAR_NAME", "2019-11-21")
env.date("VAR_NAME", "2019-11-2")
env.date("VAR_NAME", "2019-3-2")

or times:

env.time("VAR_NAME", "13:13:13.000123")
env.time("VAR_NAME", "13:13:13.123")
env.time("VAR_NAME", "13:13:13")
env.time("VAR_NAME", "13:13")

Iterables

It’s additionally possible to consume a string and cast it to a sequence etc:

env.tuple("VAR_NAME", "123,4356,235") == ("123", "4356", "235")
env.list("VAR_NAME", "123,4356,235") == ["123", "4356", "235"]
env.set("VAR_NAME", "123,4356,235") == {"123", "4356", "235"}
env.frozenset("VAR_NAME", "123,4356,235") == {"123", "4356", "235"}
env.dict("VAR_NAME", "a=1, b=2, c=3") == {"a": "1", "b": "2", "c": "3"}

Commas are treated as delimiters, and may optionally have a single space after each one.

Leading python-iterable characters are dropped if they are present from both sides, and their python type is ignored:

env.tuple("VAR_NAME", "[123, 4356, 235]") == ("123", "4356", "235")
env.tuple("VAR_NAME", "(123, 4356, 235)") == ("123", "4356", "235")
env.tuple("VAR_NAME", "{123, 4356, 235}") == ("123", "4356", "235")

Casting on iterables

Using any of env.tuple, env.list, env.set, env.frozenset, or env.dict allows each parsed value to be validated and optionally cast, with the caveat that the iterable is homogenous (that is, everything can be converted to an int or a uuid or whatever)

Each value may be cast to any of the non-iterable methods documented above, by using env.ensure.methodname instead of env.methodname, for example:

env.tuple("VAR_NAME", "123,4356,235", converter=env.ensure.int) == (123, 4356, 235)
env.set("VAR_NAME", "123,4356,235", converter=env.ensure.hex) == {"123", "4356", "235"}
env.list("VAR_NAME", "a@b.com, b@c.com, def@ghi", env.ensure.email) == ['a@b.com', 'b@c.com', 'def@ghi']

env.dict is slightly special in that it has arguments for key_converter and value_converter so that keys can have a different type to values. Both must still be homogenous:

env.dict("VAR_NAME", "a=1, b=2, c=3", key_converter=env.ensure.hex, value_converter=env.ensure.int) == {'a': 1, 'c': 3, 'b': 2}

Handling errors

Failing to successfully convert (or just validate) the value (whether from the environment or from the fallback) immediately halts execution by raising EnvironmentCastError which is a subclass of ValueError.

Failing to provide a string for a default/fallback value will raise EnvironmentDefaultError which is also a subclass of ValueError.

To catch any anticipated error then, is to:

try:
    ...
except (EnvironmentCastError, EnvironmentDefaultError) as e:
    ...

Checking for existence

To find out if an environment variable is set, regardless of it’s value, you can use normal in testing:

if "MY_KEY" in env:
    do_something_special()

which allows you to change behaviour based on seeing certain variables in the running environment.

Tracking the requests

Every access of an Environment (eg: the default env) keeps an internal log of the key requested + whether or not it was found and used in the environment.

These are available under env.used and env.fallbacks but may be accessed together by iterating over the Environment in question, where each iteration will yield a 3-tuple of:

  • environment variable name requested

  • the default or fallback value

  • a bool of whether or not the environment variable was used or whether the fallback was. True if found in the environment, False if the fallback was used.

For example, to output everything, you might do:

from enviable import env, Environment
import sys
env.int("TEST", "4")
myenv = Environment({"TESTING": "1"})
myenv.bool("TESTING", "0")
if __name__ == "__main__":
    for env_var_name, env_var_example, was_read_from_env in env:
        if was_read_from_env is True:
            sys.stdout.write("{} was in the environment\n".format(env_var_name))
        else:
            sys.stdout.write("{} was NOT in the environment, used default value of {}\n".format(env_var_name, env_var_example))

Note that in the above scenario, because env and myenv are different instances with their own individual tracking, the request for TESTING will not output, but TEST will.

Running the tests

Given a copy of the file enviable.py you ought to be able to do either of the following:

$ python enviable.py
$ python -m enviable

and see the output of the various tests I’ve bothered with. If mypy is installed, it will also type-check the file.

TODO

  • More tests

The license

It’s FreeBSD. There’s should be a LICENSE file in the root of the repository, and in any archives.


Copyright (c) 2019, Keryn Knight All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

  1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.

  2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.


Change history for enviable

0.2.2

  • Bugfix: the typing module is no longer a hard-dependency, because whilst you have it on py3, you only might have it in a py2 environment. See GitHub issue 2

0.2.1

  • Feature: Testing if a key exists in the environment can now be done using "KEYNAME" in env

  • Bugfix: using env.one_of(key, default="", choices="") without providing a custom converter no longer throws a TypeError

  • Misc: Updated type hints across the module to keep mypy happy, and me deeply unhappy.

  • Misc: Before executing tests when doing python -m enviable, try and run mypy

0.2.0

  • Bugfix: Resolved issue where by default it wasn’t stripping leading/trailing whitespace from incoming strings

  • Feature: Added new convienience method env.one_of(key, default="", choices="", converter=None) for providing a constrainted list of options which are valid.

  • Misc: Updated the default converter for iterables to be a partial

  • Bugfix: Resolved issue where str() and repr() for an Environment would leave duplicate values in if the same variable is searched for multiple times with different example/default values.

0.1.0

  • Initial export

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

enviable-0.2.2.tar.gz (19.9 kB view hashes)

Uploaded Source

Built Distribution

enviable-0.2.2-py2.py3-none-any.whl (14.8 kB view hashes)

Uploaded Python 2 Python 3

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