Skip to main content

Module properties for Python

Project description

Description

This package intends to offer a mechanism to add properties to your modules. The primary use-case of this functionality is to allow for the deferred execution of expensive-to-compute module/package level globals without needing to explicitly call a function.

Generally speaking, this module has two APIs that offer this functionality. You can use either of them or both of them, and everything will work more or less as you expect it to, as long as you follow the rules.

Note that this package supports the use of basically any descriptor at the module-level, not just properties.

Usage

The first method of using the package supports basically any descriptor being used as a decorator in the standard way. We show the use of property for the sake of brevity:

@property
def module_property(mod):
    '''mod is the module that this property was defined in'''

import mprop; mprop.init()

Because remembering to put the import/init call at the bottom of a module can be annoying, we’ve got a special decorator that works just like the property object, but handles the mprop.init() call for you:

from mprop import mproperty

@mproperty
def module_property2(mod):
    '''I work exactly the same as the earlier module property, but you
    don't need to make a subsequent call to ``mprop.init()``'''

Regardless of which method you use, if the name of your module is mod those that import the module can access the properties as normal:

# example.py
from mprop import mproperty

@mproperty
def prop(mod):
    return "I was called!"

def fcn():
    # I can access the property via:
    print _pmodule.prop

# test.py
import example

# the below should print "I was called!"
print example.prop
# this should also print "I was called!"
example.fcn()

Referencing properties from within the module

After initialization, your code may want to reference the global properties. If you try to access the properties directly, you will get a NameError unless you locally aliased the value, the initialization has not completed, or unless someone else injected a value with that name into the globals.

If you would like to directly access properties in the module, from within the module, you must reference them relative to the newly generated module. This is available from within your functions defined in the module via _pmodule, which is the “property-enhanced module” that offers property access.

If you find it necessary to require access to the original module object (which doesn’t support properties), you can access _module from the global namspace.

How it works

The short version: we replace the standard Python module instance in sys.modules during module import, which allows us to ensure that the module is replaced everywhere it is used. We perform some magic to ensure that everything available in the original module is available in the replacement module (the replacement module’s __dict__ is the module’s globals), and we post-process everything in the module’s global namespace to pull out descriptors as necessary.

As of February 1, 2019, in order to offer proper @mproperty.setter/deleter support, we’ve started initializing the module after import via the system profiler. More specifically, any time we get a call for @mproperty, we see if we’ve installed a system profiler to clean up after this module. In CPython via the sys.setprofile() handler, we look for return calls from module creation. These auto-nest, and auto-remove themselves, so by the time your final mprop-using module is done initializing, our profilers are removed.

We used to inject a special __getattribute__ method that auto-initializes on first access, but it was being triggered in cases where I didn’t expect.

Python magic that goes on

  • Using sys.getframe() to pull calling frame objects to get module globals and subsequently the module object itself

  • Replace sys.modules entries

  • Replace an instance __dict__

  • Assign descriptors during runtime by modifying the base class of the replacement module

  • Use the system profiler to discover when a module has finished loading, to start initialization

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

mprop-0.17.0.tar.gz (14.3 kB view details)

Uploaded Source

File details

Details for the file mprop-0.17.0.tar.gz.

File metadata

  • Download URL: mprop-0.17.0.tar.gz
  • Upload date:
  • Size: 14.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.1.1 CPython/3.11.0rc1

File hashes

Hashes for mprop-0.17.0.tar.gz
Algorithm Hash digest
SHA256 e14597c821481ee569aa80ec79a71345e6a627cb8b1e3e2038450fe2128eabd2
MD5 89201c1be95e3a06d5242d801409cc31
BLAKE2b-256 41f2c949e0dae816dc173b1fa30ef26588bdcd94fd20478831d274a44f691a07

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page