Simplifies working with configs and params.
This will install the latest version of the major version 1
.
It's safe enough due to followed by the project Semantic Versioning paradigm.
pip install "simputils-config~=1.0"
Warning
Developer of this project has nothing to do with simputils
package of pypi.
And installation of both might cause broken code.
On the end of this project, the namespace simputils
is made "shareable".
But the structure of another developer's package is not designed in such way
Disclaimer about that you can find here Potential package collision 2024
Class simputils.config.models.ConfigStore
is the keystone of the library.
Object of it represents config, that could be used to sequentially apply different sets of key-value
pairs from different sources (other dicts, files of different kinds, environment variables, etc.).
Object will store the latest values of the applied sets, where the earliest applied values could be overriden by the following ones. Basically, just a dictionary.
But besides being like a dictionary, it records the "history" of applied configs, so it's easier to debug and control from where which value came from.
The major purpose of the functionality comes from using multiple files as sources for your config, but it still allows to apply sets directly from the code.
Project follows Semantic Versioning standard.
No breaking changes suppose to be within minor
and patch
versions of the same major
version.
Exception is only if there is a bug, that gets fixed.
Caution
ConfigStore
object is behaving like a dict
, so if you need to check if
the variable with this object is None, always check it like if conf is None:
,
and never check it like if not conf:
!
The check is implicit, so when you use not
or simple boolean check, it will check if the object
is empty (does not contain any value)
Note
To check if ConfigStore
object contains at least one key-value pair,
you can use simple if conf:
When working with files, keep in mind that the only supported files are .yml
/.yaml
, .env
and .json
.
If you need support for other types, you will have to implement your custom handler(s) for those file-types.
- Changelog - Please make sure you check it for new features and changes
- The overall example
- Working with enums and annotations
- Config Object Style Access
- Preprocessing and filtering
- Working with
ConfigHub
(Quick Start, recommended way to work with the configs) - Working with
ConfigStore
- Config Merging Strategies
There are 3 config modifiers that could be applied: preprocessor
, filter
and annotation type-casting
[!INFO] The order of config modifiers applied to each config:
preprocessor
filter
- Annotation type-casting
Aggregation from multiple sources (you can specify any number of sources).
Important
Keep in mind that order matters, keys/values from the latest might override already specified
From files:
from simputils.config.components import ConfigHub
conf = ConfigHub.aggregate(
"config-1.yml",
"config-2.yml",
# ...
)
print(conf, conf.applied_confs)
From dictionaries:
from simputils.config.components import ConfigHub
conf = ConfigHub.aggregate(
{"key1": "val1", "key2": "val2"},
{"key2": "redefined val2", "key3": "val3"},
# ...
)
print(conf, conf.applied_confs)
Accessing values:
# Accessing values by name, if does not exist - None is returned
conf["key1"]
conf["key2"]
# Accessing values by name, if does not exist or None - `default` parameter value is returned
conf.get("key1", "My Default Value")
conf.get("key2")
# Iterating as a dictionary
for key, val in conf.items():
print(key, val)
import os
from typing import Annotated
from simputils.config.base import simputils_pp
from simputils.config.components import ConfigHub
from simputils.config.generic import BasicConfigEnum
from simputils.config.models import ConfigStore, AnnotatedConfigData
class MyEnum(BasicConfigEnum):
# Annotated without and with default values set
MY_E_KEY_1: Annotated[str, AnnotatedConfigData()] = "my-e-key-1"
MY_E_KEY_2: Annotated[str, AnnotatedConfigData(
default=3.1415
)] = "my-e-key-2"
# Non-annotated, so they will be None by default
MY_E_KEY_3 = "my-e-key-3"
MY_E_KEY_4 = "my-e-key-4"
MY_E_KEY_5 = "my-e-key-5"
# Some of them used in `app-conf.yml`
MY_FIRST_VAL = "val-1"
MY_SECOND_VAL = "VAL_2"
# Will be taken from os.environ,
# all other os.environ values will be excluded
ENV_USER_NAME = "USER"
conf = ConfigHub.aggregate(
"tests/data/config-1.yml",
os.environ,
target=ConfigStore(
MyEnum.defaults(),
preprocessor=simputils_pp,
filter=True
),
)
print("conf: ", conf)
conf: {
'MY_E_KEY_1': None,
'MY_E_KEY_2': 3.1415,
'MY_E_KEY_3': None,
'MY_E_KEY_4': None,
'MY_E_KEY_5': None,
'VAL_1': 'My conf value 1',
'VAL_2': 'My conf value 2',
'USER': 'ivan'
}
Enum
keys are supported out of the box, and argparse.Namespace
could be used for ConfigStore
Note
BasicConfigEnum
is used for convenience.
And it's suggested way, it allows to use True
as a filter key
You still can use Enums without that class,
just make sure that enum is inherited from str, Enum
in that order!
from argparse import ArgumentParser
from simputils.config.base import simputils_pp
from simputils.config.components import ConfigHub
from simputils.config.models import ConfigStore
from simputils.config.generic import BasicConfigEnum
args_parser = ArgumentParser()
args_parser.add_argument("--name", "-n", default="PandaHugMonster")
args_parser.add_argument("--age", default="33")
args = args_parser.parse_args(["--name", "Oldie", "--age", "34"])
class MyEnum(BasicConfigEnum):
MY_1 = "key-1"
MY_2 = "key-2"
MY_3 = "key-3"
NAME = "name"
AGE = "age"
c = ConfigHub.aggregate(
{MyEnum.MY_2: "new val 2", "test": "test"},
args,
target=ConfigStore(
MyEnum.defaults(),
preprocessor=simputils_pp,
filter=True
)
)
import os
from simputils.config.components import ConfigHub
# Sequence of files/values matter!
app_conf = ConfigHub.aggregate(
"data/app-conf.yml",
# This one does not exist, and used only for local redefinitions of developer or stage
"data/app-conf-local.yml",
)
# Sequence of files/values matter!
app_env_vars = ConfigHub.aggregate(
"data/production.env",
# This one does not exist, and used only for local redefinitions of developer or stage
"data/production-local.env",
# This one does not exist, and used only for local redefinitions of developer or stage
".env",
# Environment Variables from OS
os.environ,
)
print("App Conf: ", app_conf)
print("App EnvVars: ", app_env_vars)
App Conf: {
'val-1': 'My conf value 1',
'val-2': 'My conf value 2',
'val-3': 'My conf value 3'
}
App EnvVars: {
'APP_MY_ENV_VAR_1': '1',
'APP_MY_ENV_VAR_2': '2',
'APP_MY_ENV_VAR_3': '3',
...(values from OS env-vars are here)
}