pyfarm.core.config module

Configuration Object

Basic module used for reading configuration data into PyFarm in various forms.

 set of values which will return a True boolean value from read_env_bool()
 set of values which will return a False boolean value from read_env_bool()
class pyfarm.core.config.Configuration(name, version=None, cwd=None)

Bases: dict

Main object responsible for finding, loading, and merging configuration data. By default this class does nothing until load() is called. Once this method is called Configuration class will populate itself with data loaded from the configuration files. The configuration files themselves can be loaded from multiple location depending on the system’s setup. For example on Linux you might end up attempting to load these files for pyfarm.agent v1.2.3:

Override paths set by DEFAULT_ENVIRONMENT_PATH_VARIABLE. By default this path will not be set, this is only an example.

  • /tmp/pyfarm/agent/1.2.3/agent.yml
  • /tmp/pyfarm/agent/1.2/agent.yml
  • /tmp/pyfarm/agent/1/agent.yml
  • /tmp/pyfarm/agent/agent.yml

Paths relative to the current working directory or the directory provided to cwd when Configuration was instanced.

  • etc/pyfarm/agent/1.2.3/agent.yml
  • etc/pyfarm/agent/1.2/agent.yml
  • etc/pyfarm/agent/1/agent.yml
  • etc/pyfarm/agent/agent.yml

User’s home directory

  • ~/.pyfarm/agent/1.2.3/agent.yml
  • ~/.pyfarm/agent/1.2/agent.yml
  • ~/.pyfarm/agent/1/agent.yml
  • ~/.pyfarm/agent/agent.yml

System level paths

  • /etc/pyfarm/agent/1.2.3/agent.yml
  • /etc/pyfarm/agent/1.2/agent.yml
  • /etc/pyfarm/agent/1/agent.yml
  • /etc/pyfarm/agent/agent.yml

Finally, if we don’t locate a configuration file in any of the above paths we’ll use the file which was installed along side the source code.

Configuration will only attempt to load data from files which exist on the file system when load() is called. If multiple files exist the data will be loaded from each file with the successive data overwriting the value from the previously loaded configuration file. So if you have two files containing the same data:

  • /etc/pyfarm/agent/agent.yml

        a: 0
    foo: 1
    bar: true
  • etc/pyfarm/agent/1.2.3/agent.yml

        a: 1
        b: 1
    foo: 0

You’ll end up with a single merged configuration. Please note that the only keys which will be merged in the configuration are the env key. Configuration files are meant to store simple data and while it can be used to store more complicate data it won’t merge any other data structures.

    a: 1
    b: 1
foo: 0
bar: true
  • DEFAULT_SYSTEM_ROOT (string) –

    The system level directory that we should look for configuration files in. This path is platform dependent:

    The value built here will be copied onto the instance as system_root

  • DEFAULT_USER_ROOT (string) –

    The user level directory that we should look for configuration files in. This path is platform dependent:

    The value built here will be copied onto the instance as user_root

  • DEFAULT_FILE_EXTENSION (string) – The default file extension of the configuration files. This will default to .yml and will be copied to file_extension when the class is instanced.
  • DEFAULT_LOCAL_DIRECTORY_NAME (string) – A directory local to the current process which we should search for configuration files in. This will default to etc and will be copied to local_dir when the class is instanced.
  • DEFAULT_PARENT_APPLICATION_NAME (string) – The base name of the parent application. This used used to build child directories and will default to pyfarm.
  • DEFAULT_ENVIRONMENT_PATH_VARIABLE (string) – A environment variable to search for a configuration path in. The value defined here, which defaults to PYFARM_CONFIG_ROOT, will be read from the environment when Configuration is instanced. This allows for an non-standard configuration location to be loaded first for testing forced-override of the configuration.
  • DEFAULT_TEMP_DIRECTORY_ROOT – The directory which will store any temporary files.
  • name (string) – The name of the configuration itself, typically ‘master’ or ‘agent’. This may also be the name of a package such as ‘pyfarm.agent’. When the package name is provided we can usually automatically determine the version number.
  • version (string) – The version the version of the program running.
  • cwd (string) – The current working directory to construct the local path from. If not provided then we’ll use os.getcwd() to determine the current working directory.

Performs variable expansion for value. This method is run when a string value is returned from get() or __getitem__(). The default behavior of this method is to recursively expand variables using sources in the following order:

  • The environment, os.environ
  • The environment (from the configuration), env
  • Other values in the configuration
  • ~ to the user’s home directory

For example, the following configuration:

foo: foo
bar: bar
foobar: $foo/$bar
path: ~/$foobar/$TEST

Would result in the following assuming $TEST is an environment variable set to somevalue and the current user’s name is user:

    "foo": "foo",
    "bar": "bar",
    "foobar": "foo/bar",
    "path": "/home/user/foo/bar/somevalue"
DEFAULT_USER_ROOT = '/home/docs'
directories(validate=True, unversioned_only=False)

Returns a list of platform dependent directories which may contain configuration files.

  • validate (bool) – When True this method will only return directories which exist on disk.
  • unversioned_only (bool) – When True this method will only return versionless directories instead of both versionless and versioned directories.
files(validate=True, unversioned_only=False)

Returns a list of configuration files.

  • validate (bool) –

    When True this method will only return files which exist on disk.


    This method calls directories() and will be passed the value that is provided to validate here.

  • unversioned_only (bool) – See the keyword documentation for unversioned_only in directories`()
get(key, default=None)

Overrides dict.get() to provide internal variable expansion through _expandvars().


Loads data from the configuration files. Any data present in the env key in the configuration files will update environment

Parameters:environment (dict) – A dictionary to load data in the env key from the configuration files into. This would typically be set to os.environ so the environment itself could be updated.

Splits self.version into a tuple of individual versions. For example 1.2.3 would be split into ['1', '1.2', '1.2.3']

pyfarm.core.config.read_env(envvar, default=<object object>, warn_if_unset=False, eval_literal=False, raise_eval_exception=True, log_result=True, desc=None, log_defaults=False)

Lookup and evaluate an environment variable.

  • envvar (string) – The environment variable to lookup in os.environ
  • default (object) – Alternate value to return if envvar is not present. If this is instead set to NOTSET then an exception will be raised if envvar is not found.
  • warn_if_unset (bool) – If True, log a warning if the value being returned is the same as default
  • eval_literal – if True, run literal_eval() on the value retrieved from the environment
  • raise_eval_exception (bool) – If True and we failed to parse envvar with literal_eval() then raise a EnvironmentKeyError
  • log_result (bool) – If True, log the query and result to INFO. If False, only log the query itself to DEBUG. This keyword mainly exists so environment variables such as PYFARM_SECRET or PYFARM_DATABASE_URI stay out of log files.
  • desc (string) – Describes the purpose of the value being returned. This may also be read in at the time the documentation is built.
  • log_defaults (bool) – If False, queries for envvars that have not actually been set and will just return default will not be logged
pyfarm.core.config.read_env_bool(*args, **kwargs)

Wrapper around read_env() which converts environment variables to boolean values. Please see the documentation for read_env() for additional information on exceptions and input arguments.

  • AssertionError – raised if a default value is not provided
  • TypeError – raised if the environment variable found was a string and could not be converted to a boolean.
pyfarm.core.config.read_env_number(*args, **kwargs)

Wrapper around read_env() which will read a numerical value from an environment variable. Please see the documentation for read_env() for additional information on exceptions and input arguments.

Raises:TypeError – raised if we either failed to convert the value from the environment variable or the value was not a float, integer, or long
pyfarm.core.config.read_env_strict_number(*args, **kwargs)

Strict version of read_env_number() which will only return an integer


number_type – the type of number(s) this function must return

  • AssertionError – raised if the number_type keyword is not provided (required to check the type on output)
  • TypeError – raised if the type of the result is not an instance of number_type