Skip to content

Latest commit

 

History

History
301 lines (252 loc) · 10.7 KB

README.md

File metadata and controls

301 lines (252 loc) · 10.7 KB
Raw

pconf

Build Status Code Coverage PyPi Version License Py Versions Downloads Known Vulnerabilities Code style: black

Hierarchical python configuration with files, environment variables, command-line arguments.

Example

from pconf import Pconf
import json

"""
Setup pconf config source hierarchy as:
  1. Environment variables
  2. A JSON file located at 'path/to/config.json'
"""
Pconf.env()
Pconf.file('path/to/config.json', encoding='json')

# Get all the config values parsed from the sources
config = Pconf.get()

# Just print everything nicely
print json.dumps(config, sort_keys=True, indent=4)

Run the above script:

python example.py

The output should be something like this:

{
    "HOSTNAME": "bb30700d22d8",
    "TERM": "xterm",
    "PATH": "/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
    "PWD": "/",
    "SHLVL": "1",
    "HOME": "/root",
    "no_proxy": "*.local, 169.254/16",
    "_": "/usr/bin/env",
    "example": {
        "another": "stuff",
        "key": "value"
    }
}

Hierarchical configuration

Pconf is designed to be used with multiple sources of configuration values with the user being able define the priority of each of these as a hierarchy. The supported sources described below can be setup in any order without any hardcoded defaults. Priority meaning if a configuration key appears in multiple sources the value from the sources higher up in the hierarchy takes precedence. The order in which the sources are attached defines the priority in the hierarchy, source attached first take precedence.

The available sources (more details about the below) in a sensible order:

  1. overrides - Loads data passed to it
  2. argv - Parses command line arguments to the process
  3. env - Parses environment variables
  4. file - Parses config files of various formats
  5. defaults - Loads data passed to it

Config sources

Defaults, overrides

These two sources are essentially the same, pass a dict to when attaching and they will return that when queried.

Pconf.overrides({'key': 'override_value'})
Pconf.defaults({'key': 'default_value'})

Very simple, as the name suggests these are to allow the user to set defaults and override whatever value.

Argv

Responsible for loading values parsed from command line arguments passed to the process. Parameters passed to the process, but not described to be parsed as below are ignored.

Parsed arguments can be defined with the following parameters:

  • name: the long name of the argument
  • short_name: the optional short name of the argument
  • type: the optional type of the argument
  • help: the optional help text for the argument
Pconf.argv('--test_argument')
Pconf.argv('--privileged', type=bool)
Pconf.argv('--threads', short_name='-c', type=int)
Pconf.argv('--verbose', short_name='-v', type=bool, help='Run in verbose mode')

These could be used like:

python example.py --test_argument=hello_world -v --threads 4

Lists as arguments

Lists can be passed in two ways as arguments

  • literal list
Pconf.argv('--list', type=list)
python example.py --list="['item1', 'item2']"
{'list': ['item1', 'item2']}
  • repeated arguments:
Pconf.argv('--list', type='repeated_list')
python example.py --list=item1 --list=item2
{'list': ['item1', 'item2']}

Using alongside argparse

If you want to use argparse for not only config values it is possible to integrate the values from there with the structure uesd by Pconf:

parsed_args = argparse.ArgumentParser(...).parse_args()
Pconf.overrides(vars(parsed_args))

Env

Responsible for loading values parsesd from os.environ into the configuration hierarchy.

# Just load all the variables available for the process
Pconf.env()

# A separator can be specified for nested keys
Pconf.env(separator='__')
# This turns the 'log__file=/log' env variable into the `{'log': {'file': '/log'}}` dict

# Available variables can be whitelisted
Pconf.env(whitelist=['only', 'load', 'variables', 'listed', 'here'])

# A regular expression can be specified for matching keys also
# Keys matched by this expression are considered whitelisted
Pconf.env(match='^REGEX.*')

# The contents of the variables can be parsed as well-known types, in case parsing fails the variable remains as a string
Pconf.env(parse_values=True)

# The variable names can be converted to lowercase to match other variable names.
Pconf.env(to_lower=True)

# Convert all underscores in the name to dashes, this takes place after separation via the separator option.
Pconf.env(convert_underscores=True)

# Handle docker secrets, described in detail below
Pconf.env(docker_secrets=['MYSQL_ROOT_PASSWORD_FILE', 'MYSQL_PASSWORD_FILE']

# Use all at once
Pconf.env(separator='__',
          match='whatever_matches_this_will_be_whitelisted',
          whitelist=['whatever', 'doesnt', 'match', 'but', 'is', 'whitelisted', 'gets', 'loaded', 'too'],
          parse_values=True,
          to_lower=True,
          convert_underscores=True)

Docker secret handling

As described in https://docs.docker.com/v17.12/engine/swarm/secrets/#build-support-for-docker-secrets-into-your-images, when using the docker_secrets parameter as above, given the following:

  • A docker secret called db_root_password containing the value secret-password
  • The secret passed to the environment in which Pconf is running
  • The MYSQL_ROOT_PASSWORD_FILE environment variable set to /run/secrets/db_root_password The result of
Pconf.env(docker_secrets=['MYSQL_ROOT_PASSWORD_FILE'])
Pconf.get()

will be

{'MYSQL_ROOT_PASSWORD': 'secret-password'}

Notice that the environment name passed was stripped from the _FILE postfix.

Note: this behaviour is not tied specifically to docker secrets, it will read files specified in the variable without regard to its path, will simply drop the value if the file is not found or the variable name does not end with _FILE.

Combining this with the other parameters: the files are read after regex matching and whitelisting, but before parsing and conversions.

File

Responsible for loading values parsed from a given file into the configuration hierarchy. If the file does not exist the result will be empty and no error is thrown.

By default tries to parse file contents as literal python variables, use the encoding parameter to set the file format/encoding.

"""
`/path/to/literal` contents:
{'this': 'is_a_literal_python_dict'}
"""
Pconf.file('/path/to/literal')

Built-in encodings:

These are the built-in supported encodings, that can be passed as the encoding parameter to the function. All the example files below result in the same data structure after parsing.

  • ini 
    """
`/path/to/config.ini` contents:
[example]
key: value
another=stuff
"""
Pconf.file('/path/to/config.ini', encoding='ini')
    Ini files have to have sections, though the special DEFAULT section won't be included as a section header only its contents. Also ini files are quite restrictive, only basic types are supported, and most are parsed as strings. If support for more complex types is required AND it has to come from ini files, then look at custom file formats below.
  • json 
    """
`/path/to/config.json` contents:
{
    "example": {
        "key": "value",
        "another": "stuff"
    }
}
"""
Pconf.file('/path/to/config.json', encoding='json')
  • yaml 
    """
`/path/to/config.yaml` contents:
---
example:
  key: value
  another: stuff
"""
Pconf.file('/path/to/config.yaml', encoding='yaml')

Using custom file formats

To use custom encodings supply a parser along with an encoding that is not built-in. The parser is a function that expects the file contents as its argument and returns a dict containing the parsed contents.

def custom_parser(file_contents):
    return {'example': file_contents}

Pconf.file('/path/to/custom/file', encoding='example', parser=custom_parser)

Getting the config values

Use the get method to get the processed config values. The method returns all the values as a python dictionary, with nested values and all. Values can be accessed as expected from a dict.

config = Pconf.get()

print config['key']

Clearing content

There are cases when there is a need for multiple configurations, to make this work after setting up a hierarchy one should save it in a dictionary and then use the clear() method to reset Pconf's internal state.

>>> Pconf.defaults({"foo": "bar"})
>>> config = Pconf.get()
>>> config
{'foo': 'bar'}
>>> Pconf.clear()
>>> Pconf.get()
{}
>>> config
{'foo': 'bar'}

Run tests and linter

First install the dev requirements:

pip install -r requirements.txt
pip install -r requirements-dev.txt

Tests

Test are written using the standard python unittest framework. Run the tests from the repository root like so:

py.test

Linter

For code linting flake8 is used. Run the linter from the repository root like so:

flake8 --max-line-length=140 pconf tests

Contributing

Please read CONTRIBUTING.md and CODE_OF_CONDUCT.md for details on our code of conduct, and the process for submitting pull requests to us.

Versioning

We use SemVer for versioning. For the versions available, see the tags on this repository.

Authors

License

This project is licensed under the MIT License - see the LICENSE file for details

Acknowledgments