holy-diver 0.1.0a3

Config

Here’s a quick example of how to use the Config class. First create a Config object from a dictionary:

from holy_diver import Config

config_data = {
    "database": {
        "host": "localhost",
        "port": 5432,
        "credentials": {
            "user": "admin",
            "password": "secret",
        },
    }
}

config = Config.from_dict(config_data)

Access nested keys attribute-style using dot notation:

print(config.database.host)  # Output: localhost
print(config.database.port)  # Output: 5432

You can also set values using dot notation:

config.database.host = "impala.company.com"
print(config.database.host)  # Output: impala.company.com

Alternatively, you can directly look up nested keys:

print(config["database.host"])  # Output: localhost
print(config["database.port"])  # Output: 5432

And of course, you can access them the old fashioned way:

print(config["database"]["host"])  # Output: localhost
print(config["database"]["port"])  # Output: 5432

ConfigList

Here’s a quick example of how to use the ConfigList class. Items in ConfigList can be accessed using normal indexing and dot notation interchangeably. All indices can be accessed entirely with dot notation, which allows for easier handling of nested keys and data structures.

from holy_diver import ConfigList

list_data = [
    {"name": "Alice", "age": 30},
    {"name": "Bob", "age": 25}
]

config_list = ConfigList.from_list(list_data)

Access elements using indices and dot notation for nested keys:

print(config_list[0].name)  # Output: Alice
print(config_list[1].age)   # Output: 25

Or, do it all with dot notation, if you prefer:

print(config_list._0.name) # Output: Alice
print(config_list._1.age)  # Output: 25

The leading underscore allows numeric indices to be accessed as attributes. The leading underscore is always required for attribute access, but is optional in other contexts. You can see all the nested keys using the deep_keys() method, which shows the leading underscore for numeric indices:

print(config_list.deep_keys())
# Output: ['_0', '_1', '_0.name', '_0.age', '_1.name', '_1.age']

You can also look up nested keys directly:

print(config_list["_0.name"]) # Output: Alice
print(config_list["_1.age"])  # Output: 25

# It also works without the underscore
print(config_list["0.name"]) # Output: Alice
print(config_list["1.age"])  # Output: 25

Loading from a Configuration File

You can load a configuration file in YAML format using the Config.from_yaml() method:

from holy_diver import Config

config = Config.from_yaml("config.yaml")

Loading a JSON file works in much the same way:

from holy_diver import Config

config = Config.from_json("config.json")

Alternative Constructors

It’s generally recommended to use one of the from_*() constructors (e.g. from_dict(), from_yaml()) to create either a Config or ConfigList, because these class methods automatically convert nested dictionaries and lists to manager classes. It shouldn’t affect the functionality much if you use the main constructor, but it may cost you a few milliseconds of processing time down the road, as more conversions must be performed on the fly.

Writing to a Configuration File

You can dump the configuration in various formats: YAML, JSON, and TOML. Simply use the corresponding to_*() method (e.g. to_yaml(), to_json()) and supply a path. Note that ConfigList objects can only be dumped to YAML and JSON.

Converting and Deconverting

If you want to, you can convert the entire hierarchy to nested managers using the convert() method. This is done automatically when using the from_*() constructors, but if you’ve used the main constructor or added some keys and values (an odd thing to do), you might want to obtain a converted copy of the hierarchy. Again, this has a barely noticeable effect on the functionality. Alternatively, you can deconvert the hierarchy to nested dicts and lists using the deconvert() method. This is useful if you want the configuration data in vanilla Python data structures for serialization.

from holy_diver import Config

config_data = {
    "database": {
        "host": "localhost",
        "port": 5432,
        "credentials": {
            "user": "admin",
            "password": "secret",
        },
    }
}

config = Config(config_data) # Create a manager using main constructor
converted = config.convert() # Convert to nested managers
deconverted = converted.deconvert() # Deconvert to nested dicts and lists

# Access nested keys
print(config.database.host)  # Output: localhost
print(converted.database.host)  # Output: localhost
print(deconverted["database"]["host"])  # Output: localhost

Setting Defaults

You can set default values for keys that may not be present in the configuration data. Simply pass the defaults keyword argument to any of the Config constructors. This argument should be a dictionary of default values. If a key is not present in the configuration data, the default value will be used instead. The user configuration is recursively merged with the defaults to ensure that nested keys are handled properly.

from holy_diver import Config

default_config = {
    "database": {
        "host": "impala.megacorp.com", # Will be overridden
        "port": 21050, # Will be overridden
        "auth_method": "LDAP", # Not present in the config data
    }
}
config_data = {"database": {"host": "localhost", "port": 5432}}

config = Config.from_dict(config_data, defaults=default_config)

print(config.database.host)  # Output: localhost
print(config.database.port) # Output: 5432
print(config.database.auth_method)  # Output: LDAP

Checking for Required Keys

One of the nice features of Config is that it allows you to check for the presence of required keys. This is especially useful because it works for nested keys using dot notation.

from holy_diver import Config

config_data = {
    "database": {
        "host": "localhost",
        "port": 5432,
        "credentials": {
            "user": "admin",
            "password": "secret",
        },
    }
}

required_keys = ["database.host", "database.credentials.user", "database.auth_method"]

config = Config.from_dict(config_data) # Create a manager

config.check_required_keys(required_keys, if_missing="raise")
# Output: KeyError: Configuration is missing required keys: ['database.auth_method']

Raise a warning instead of an exception by passing if_missing="warn":

missing_keys = config.check_required_keys(required_keys, if_missing="warn")
# Output: UserWarning: Configuration is missing required keys: ['database.auth_method']
print(missing_keys) # Output: ["database.auth_method"]

Or, quietly get a list of missing keys by passing if_missing="return":

missing_keys = config.check_required_keys(required_keys, if_missing="return")
print(missing_keys) # Output: ["database.auth_method"]

You can also check for required keys by passing required_keys to any of the Config constructors.

config = Config.from_dict(config_data, required_keys=required_keys)
# Output: KeyError: Configuration is missing required keys: ['database.auth_method']

Credits

This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.