Skip to main content

This module provides support for easy addition of private attributes inside your custom objects, which are totally unreachable from outside the class definition, as in C++ 'private' clause.

Project description

private-attrs

PyPI PyPI - Downloads PyPI - Python Version PyPI - Status

GitHub code size in bytes GitHub last commit Build Status Maintenance

This little library, consisting of a single module, provides support for easy addition of truly private attributes inside classes, which are totally unreachable from outside the class definition, as in C++ private clause.

Table of contents

Installation

Use the package manager pip to install private-attrs.

pip3 install private-attrs

Usage

This is a simple schema on how a custom class could use private attributes.

from private_attrs import PrivateAttrs

def MyClass():
    p = PrivateAttrs()

    class MyClass:
        def __init__(self):
            p.register_instance(self)
            
        # From now on, we'll define our public attrs with 'self.attr' syntax as usual,
        # and private ones with 'p.attr' or 'p.attr_static'.

        def __del__(self):
            p.delete(self)

    MyClass.__qualname__ = 'MyClass'    

    return MyClass

MyClass = MyClass()  # override the function definition

As you can see, we first need to define the class inside a function. Outside that class, but inside the function scope, we instantiate a PrivateAttrs object.

Now, inside MyClass, if we plan to have private instance attributes, and not just static ones, it's mandatory to register, in the __init__() method, the instance by calling the register_instance() function.

Finally we return MyClass and we override the function definition.


A simple example

Let's now dive into a more complete example:

from private_attrs import PrivateAttrs

def Person():
    p = PrivateAttrs()

    class Person:
        def __init__(self, name, social_security_number):
            p.register_instance(self)
            self.name = name
            p.ssn = social_security_number

        @property
        def ssn(self):
            return p.ssn

        def __eq__(self, other):
            return self.ssn == other.ssn

        def __hash__(self):
            return hash(self.ssn)

        def __str__(self):
            return f"{self.name} - {self.ssn}"

        def __del__(self):
            p.delete(self)

    Person.__qualname__ = 'Person'

    return Person

Person = Person()

Although a person can change their name, surname or even their sex, it's really unlikely (not to say impossible) for someone to change their social security number (SSN).

That's why we store the SSN as a private attribute, safe, unmodifiable, and we can rely on it to compare whether two people are the same person.


An example with proxy=True

If we are working with the Python multiprocessing library and we want to create a class with private attributes that are accessible and modifiable from different running processes (we already know that, unlike threads, processes don't share memory space), we need to instantiate the PrivateAttrs object with the argument proxy = True.

Let's see an example:

from private_attrs import PrivateAttrs


def Person():
    p = PrivateAttrs(proxy=True)

    class Person:
        def __init__(self, name, social_security_number):
            p.register_instance(self)
            self.name = name
            p.cell_phones = p.manager.list()
            p.ssn = social_security_number

        @property
        def ssn(self):
            return p.ssn

        @property
        def cell_phones(self):
            return tuple(p.cell_phones)

        def add_cell_phone(self, phone):
            p.cell_phones.append(phone)

        def __str__(self):
            return f"{self.name} - {self.ssn} - {self.cell_phones}"

        def __del__(self):
            p.delete(self)

        def __getstate__(self):
            state = dict(self.__dict__)
            state['private'] = p.getstate(self)
            return state

        def __setstate__(self, state):
            private = state.pop('private')
            p.setstate(private, self)
            self.__dict__ = state

    Person.__qualname__ = 'Person'    

    return Person

Person = Person()

By doing this, all the private attributes that we store are automatically available in all processes, and you can modify them from anyone.

Pay particular attention to certain specific attributes that need to be instantiated using the Manager class, such as lists or dictionaries. Fortunately, there is an attached manager object in the PrivateAttrs class to simplify life for the programmer.

Also be aware of the need to define __getstate__() and __setstate__() magic methods as you see them so the class can be correctly serialized and deserialized with all its private attributes when shared between processes.

You should know that, the way we wrote this Person class, it's impossible for other processes to modify the public name attribute and make that change visible for the rest. This is because this attribute has not been instantiated with Manager.Value() nor inside a Manager.Namespace() or similar.

One possible workaround if you don't want to use the mentioned methods for storing shared simple attributes like str or int would be to make them private and then make a getter (@property) and a setter for each one. So the former Person class would look like this:

class Person:
    def __init__(self, name, social_security_number):
        p.register_instance(self)
        p.name = name
        p.cell_phones = p.manager.list()
        p.ssn = social_security_number

    @property
    def name(self):
        return p.name

    @name.setter
    def name(self, name):
        p.name = name

    @property
    def ssn(self):
        return p.ssn

    @property
    def cell_phones(self):
        return tuple(p.cell_phones)

    def add_cell_phone(self, phone):
        p.cell_phones.append(phone)

    def __str__(self):
        return f"{self.name} - {self.ssn} - {self.cell_phones}"

    def __del__(self):
        p.delete(self)

    def __getstate__(self):
        state = dict(self.__dict__)
        state['private'] = p.getstate(self)
        return state

    def __setstate__(self, state):
        private = state.pop('private')
        p.setstate(self, private)
        self.__dict__ = state

A word about compiling

If you plan to compile your program or library using a tool like Cython, Nuitka or similar, you should know that private-attrs uses the Python inspect module to provide the developer with a friendly interface, which does not behave as usual when the program is compiled to C. All you have to do to avoid compilation problems is to use the explicit private attribute getter and setter instead of the implicit declarations, only for non-static private attributes.

For example, replace this:

class Person:
    
    @property
    def name(self):
        return p.name

    @name.setter
    def name(self, name):
        p.name = name

For this

class Person:
    
    @property
    def name(self):
        return p.get_private_attr(self, 'name')

    @name.setter
    def name(self, name):
        p.set_private_attr(self, 'name', name)

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

License

PyPI - License

This library is licensed under the GNU General Public License v3 or later (GPLv3+)

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

private-attrs-1.0.1.tar.gz (18.1 kB view hashes)

Uploaded Source

Built Distribution

private_attrs-1.0.1-py3-none-any.whl (18.6 kB view hashes)

Uploaded Python 3

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