typenv

typenv

Version 0.2.0

Typed environment variable parsing for Python

Table of Contents generated with mdformat-toc

• Background
• Installing
• Usage
  • Basics
  • Supported types
  • Default values
  • Name prefixes
  • Name character set
  • Name uppercasing
  • Validation
  • Reading from a .env file
  • Dumping parsed values
• Acknowledgments

Background

Typenv does environment variable parsing with an API almost identical to the excellent environs. There are a few reasons why typenv might be preferred:

• Type annotated typecast functions: type checkers are able to understand types of parsed environment variables.
• More flexible prefix manipulation of environment variable names.
• Validation of environment variable names.
• Optional automatic uppercasing of environment variable names.
• Ability to generate a .env.example that shows expected types of environment variables.
• Less dependencies. No marshmallow required.

Installing

Installing from PyPI repository (https://pypi.org/project/typenv):

pip install typenv

Usage

Basics

Set environment variables:

export NAME='Harry Potter'
export AGE=14
export IS_WIZARD=true
export PATRONUM_SUCCESS_RATE=0.92
export BANK_BALANCE=134599.01
export LUCKY_NUMBERS=7,3,11
export EXTRA_DETAILS='{"friends": ["Hermione", "Ron"]}'
export FAVORITE_COLOR=0x7f0909

Parse the values in Python:

from typenv import Env

env = Env()

NAME = env.str("NAME")  # => "Harry Potter"
AGE = env.int("AGE")  # => 14
IS_WIZARD = env.bool("IS_WIZARD")  # => True
PATRONUM_SUCCESS_RATE = env.float("PATRONUM_SUCCESS_RATE")  # => 0.92
BANK_BALANCE = env.decimal("BANK_BALANCE")  # => decimal.Decimal("134599.01")
LUCKY_NUMBERS = env.list("LUCKY_NUMBERS", subcast=int)  # => [7, 3, 11]
EXTRA_DETAILS = env.json("EXTRA_DETAILS")  # => {"friends": ["Hermione", "Ron"]}
FAVORITE_COLOR = env.bytes("FAVORITE_COLOR", encoding="hex")  # => b"\x7f\t\t"

# Optional settings must have a default value
IS_DEATH_EATER = env.bool("IS_DEATH_EATER", default=False)  # => False

Supported types

The types supported by typenv are:

• env.str
• env.int
• env.bool
• env.float
• env.decimal
• env.list
  • Takes a subcast keyword argument for casting list items to one of str, int , bool, float or decimal.Decimal
• env.json
• env.bytes
  • Takes an encoding keyword argument for indicating how the bytes are encoded. For now only hex is supported.

Default values

Normally, if an environment variable is not found, typenv raises an exception. If a default value is provided, however, that will be returned instead of raising.

from typenv import Env

env = Env()

BOOL = env.bool("NON_EXISTING_NAME", default=False)  # => False
LIST = env.list("NON_EXISTING_NAME", default=["a", "b"])  # => ["a", "b"]
OPTIONAL_INT = env.int("NON_EXISTING_NAME", default=None)  # => None

Name prefixes

export FLASK_HOST=127.0.0.1
export FLASK_PORT=44144

from typenv import Env

env = Env()

# Explicitly prefixing variable names works, but repeats itself
# (especially given more environment variables and nested prefixes).
HOST = env.str("FLASK_HOST")  # => "127.0.0.1"
PORT = env.int("FLASK_PORT")  # => 44144

# This reads the same variables as above, and can be a nice way of
# reducing repetition and expressing structure. Note that it is possible
# to have nested `with` statements.
with env.prefixed("FLASK_"):
    HOST = env.str("HOST")  # => "127.0.0.1"
    PORT = env.int("PORT")  # => 44144

# For more control, one can mutate `env.prefix` (of type list[str])
# directly. Note that if an exception occurs reading the environment
# variables, then `env.prefix` will not be reset to its initial value,
# which is something that the `with` statement would take care of.
env.prefix.append("FLASK_")
HOST = env.str("HOST")  # => "127.0.0.1"
PORT = env.int("PORT")  # => 44144
env.prefix.pop()

Name character set

Typenv validates environment variable names. By default, the set of allowed characters includes upper case ASCII letters, digits and the underscore (_).

The set of allowed characters can be configured:

from typenv import Env

env = Env(allowed_chars="ABCDEFGHIJKLMNOPQRSTUVWXYZ")

Name uppercasing

export UPPER_CASE_NAME=true

from typenv import Env

# Environment variable names in type cast methods will automatically be upper
# cased when `upper=True` is set here.
env = Env(upper=True)

NAME = env.bool("upper_casE_Name")

Validation

export NAME='Harry Potter'
export AGE=14

from typenv import Env

env = Env()

# A single validator function
NAME = env.str("NAME", validate=lambda n: n.startswith("Harry"))

# A validator function can signal error by raising an exception
def is_positive(num):
    if num <= 0:
        raise Exception("Number is not positive")


# A validator function can alternatively return `False` to signal an error
def is_less_than_thousand(num):
    if num >= 1000:
        return False
    return True


# Multiple validator functions can be passed as an iterable of callables
AGE = env.int("AGE", validate=(is_positive, is_less_than_thousand))

Reading from a .env file

While developing, it is often useful to read environment variables from a file. Typenv supports this via the Env.read_end() method. The method will look for a file (by default) named .env in current working directory and import environment variables from it. If a file is not found, the method will walk up in the directory tree until a file is found or the root directory is reached. The method returns a boolean that is True if a file is found.

Given a .env file in current working directory with the following content

SOME_VAR='some value'

The following code will be able to read and parse the value

from typenv import Env

env = Env()
env.read_env()

SOME_VAR = env.str("SOME_VAR")  # => "some value"

Dumping parsed values

export SOME_STR=blaablaa
export SOME_INT=99

from typenv import Env, ParsedValue

env = Env()

SOME_STR = env.str("SOME_STR")
SOME_INT = env.int("SOME_INT")

assert env.dump() == {
    "SOME_INT": ParsedValue(value=99, type="int", optional=False),
    "SOME_STR": ParsedValue(value="blaablaa", type="str", optional=False),
}

Acknowledgments

The public API of this library is almost an exact copy of environs, which is based on envparse and django-environ. Credit for the interface goes to the authors of those libraries.