🕷️ Crawler Inspector

[](https://realpython.com/) - [Start Here](https://realpython.com/start-here/) - [Learn Python](https://realpython.com/python-metaclasses/) [Python Tutorials → In-depth articles and video courses](https://realpython.com/search?kind=article&kind=course&order=newest) [Learning Paths → Guided study plans for accelerated learning](https://realpython.com/learning-paths/) [Quizzes & Exercises → Check your learning progress](https://realpython.com/quizzes/) [Browse Topics → Focus on a specific area or skill level](https://realpython.com/tutorials/all/) [Community Chat → Learn with other Pythonistas](https://realpython.com/community/) [Office Hours → Live Q\&A calls with Python experts](https://realpython.com/office-hours/) [Live Courses → Live, instructor-led Python courses](https://realpython.com/live/) [Podcast → Hear what’s new in the world of Python](https://realpython.com/podcasts/rpp/) [Books → Round out your knowledge and learn offline](https://realpython.com/products/books/) [Reference → Concise definitions for common Python terms](https://realpython.com/ref/) [Code Mentor →Beta Personalized code assistance & learning tools](https://realpython.com/mentor/) [Unlock All Content →](https://realpython.com/account/join/) - [More](https://realpython.com/python-metaclasses/) [Learner Stories](https://realpython.com/learner-stories/) [Python Newsletter](https://realpython.com/newsletter/) [Python Job Board](https://www.pythonjobshq.com/) [Meet the Team](https://realpython.com/team/) [Become a Contributor](https://realpython.com/jobs/) - [Search](https://realpython.com/search "Search") - [Join](https://realpython.com/account/join/) - [Sign‑In](https://realpython.com/account/login/?next=%2Fpython-metaclasses%2F) [Browse Topics](https://realpython.com/tutorials/all/) [Guided Learning Paths](https://realpython.com/learning-paths/) [Basics](https://realpython.com/search?level=basics) [Intermediate](https://realpython.com/search?level=intermediate) [Advanced](https://realpython.com/search?level=advanced) *** [ai](https://realpython.com/tutorials/ai/) [algorithms](https://realpython.com/tutorials/algorithms/) [api](https://realpython.com/tutorials/api/) [best-practices](https://realpython.com/tutorials/best-practices/) [career](https://realpython.com/tutorials/career/) [community](https://realpython.com/tutorials/community/) [databases](https://realpython.com/tutorials/databases/) [data-science](https://realpython.com/tutorials/data-science/) [data-structures](https://realpython.com/tutorials/data-structures/) [data-viz](https://realpython.com/tutorials/data-viz/) [devops](https://realpython.com/tutorials/devops/) [django](https://realpython.com/tutorials/django/) [docker](https://realpython.com/tutorials/docker/) [editors](https://realpython.com/tutorials/editors/) [flask](https://realpython.com/tutorials/flask/) [front-end](https://realpython.com/tutorials/front-end/) [gamedev](https://realpython.com/tutorials/gamedev/) [gui](https://realpython.com/tutorials/gui/) [machine-learning](https://realpython.com/tutorials/machine-learning/) [news](https://realpython.com/tutorials/news/) [numpy](https://realpython.com/tutorials/numpy/) [projects](https://realpython.com/tutorials/projects/) [python](https://realpython.com/tutorials/python/) [stdlib](https://realpython.com/tutorials/stdlib/) [testing](https://realpython.com/tutorials/testing/) [tools](https://realpython.com/tutorials/tools/) [web-dev](https://realpython.com/tutorials/web-dev/) [web-scraping](https://realpython.com/tutorials/web-scraping/) [Table of Contents](https://realpython.com/python-metaclasses/#toc) - [Old-Style vs. New-Style Classes](https://realpython.com/python-metaclasses/#old-style-vs-new-style-classes) - [Old-Style Classes](https://realpython.com/python-metaclasses/#old-style-classes) - [New-Style Classes](https://realpython.com/python-metaclasses/#new-style-classes) - [Type and Class](https://realpython.com/python-metaclasses/#type-and-class) - [Defining a Class Dynamically](https://realpython.com/python-metaclasses/#defining-a-class-dynamically) - [Example 1](https://realpython.com/python-metaclasses/#example-1) - [Example 2](https://realpython.com/python-metaclasses/#example-2) - [Example 3](https://realpython.com/python-metaclasses/#example-3) - [Example 4](https://realpython.com/python-metaclasses/#example-4) - [Custom Metaclasses](https://realpython.com/python-metaclasses/#custom-metaclasses) - [Is This Really Necessary?](https://realpython.com/python-metaclasses/#is-this-really-necessary) - [Conclusion](https://realpython.com/python-metaclasses/#conclusion) Mark as Completed Share Recommended Course [ Metaclasses in Python 51m · 7 lessons](https://realpython.com/courses/python-metaclasses/)  # Python Metaclasses by [John Sturtz](https://realpython.com/python-metaclasses/#author) Reading time estimate 12m [22 Comments](https://realpython.com/python-metaclasses/#reader-comments) [advanced](https://realpython.com/tutorials/advanced/) [python](https://realpython.com/tutorials/python/) Mark as Completed Share Table of Contents - [Old-Style vs. New-Style Classes](https://realpython.com/python-metaclasses/#old-style-vs-new-style-classes) - [Old-Style Classes](https://realpython.com/python-metaclasses/#old-style-classes) - [New-Style Classes](https://realpython.com/python-metaclasses/#new-style-classes) - [Type and Class](https://realpython.com/python-metaclasses/#type-and-class) - [Defining a Class Dynamically](https://realpython.com/python-metaclasses/#defining-a-class-dynamically) - [Example 1](https://realpython.com/python-metaclasses/#example-1) - [Example 2](https://realpython.com/python-metaclasses/#example-2) - [Example 3](https://realpython.com/python-metaclasses/#example-3) - [Example 4](https://realpython.com/python-metaclasses/#example-4) - [Custom Metaclasses](https://realpython.com/python-metaclasses/#custom-metaclasses) - [Is This Really Necessary?](https://realpython.com/python-metaclasses/#is-this-really-necessary) - [Conclusion](https://realpython.com/python-metaclasses/#conclusion) [Remove ads](https://realpython.com/account/join/) Recommended Course [Metaclasses in Python](https://realpython.com/courses/python-metaclasses/) (51m) The term **metaprogramming** refers to the potential for a program to have knowledge of or manipulate itself. Python supports a form of metaprogramming for classes called **metaclasses**. Metaclasses are an esoteric [OOP concept](https://realpython.com/python3-object-oriented-programming/), lurking behind virtually all Python code. You are using them whether you are aware of it or not. For the most part, you don’t need to be aware of it. Most Python programmers rarely, if ever, have to think about metaclasses. When the need arises, however, Python provides a capability that not all object-oriented languages support: you can get under the hood and define custom metaclasses. The use of custom metaclasses is somewhat controversial, as suggested by the following quote from Tim Peters, the Python guru who authored the [Zen of Python](https://realpython.com/zen-of-python/): > “Metaclasses are deeper magic than 99% of users should ever worry about. If you wonder whether you need them, you don’t (the people who actually need them know with certainty that they need them, and don’t need an explanation about why).” > > — *Tim Peters* There are Pythonistas (as Python aficionados are known) who believe that you should never use custom metaclasses. That might be going a bit far, but it is probably true that custom metaclasses mostly aren’t necessary. If it isn’t pretty obvious that a problem calls for them, then it will probably be cleaner and more readable if solved in a simpler way. Still, understanding Python metaclasses is worthwhile, because it leads to a better understanding of the internals of [Python classes](https://realpython.com/python-classes/) in general. You never know: you may one day find yourself in one of those situations where you just know that a custom metaclass is what you want. **Get Notified:** Don’t miss the follow up to this tutorial—[Click here to join the Real Python Newsletter](https://realpython.com/bonus/newsletter-dont-miss-updates/) and you’ll know when the next installment comes out. ## Old-Style vs. New-Style Classes In the Python realm, a class [can be one of two varieties](https://wiki.python.org/moin/NewClassVsClassicClass). No official terminology has been decided on, so they are informally referred to as old-style and new-style classes. [Remove ads](https://realpython.com/account/join/) ### Old-Style Classes With old-style classes, class and type are not quite the same thing. An instance of an old-style class is always implemented from a single built-in type called `instance`. If `obj` is an instance of an old-style class, `obj.__class__` designates the class, but `type(obj)` is always `instance`. The following example is taken from Python 2.7: Python ``` ``` ### New-Style Classes New-style classes unify the concepts of class and type. If `obj` is an instance of a new-style class, `type(obj)` is the same as `obj.__class__`: Python ``` ``` Python ``` ``` ## Type and Class In Python 3, all classes are new-style classes. Thus, in Python 3 it is reasonable to refer to an object’s type and its class interchangeably. **Note:** In Python 2, classes are old-style by default. Prior to Python 2.2, new-style classes weren’t supported at all. From Python 2.2 onward, they can be created but must be explicitly declared as new-style. Remember that, [in Python, everything is an object.](https://web.archive.org/web/20151210024637/https://mail.python.org/pipermail/python-list/2015-June/691689.html) Classes are objects as well. As a result, a class must have a type. What is the type of a class? Consider the following: Python ``` ``` The type of `x` is class `Foo`, as you would expect. But the type of `Foo`, the class itself, is `type`. In general, the type of any new-style class is `type`. The type of the built-in classes you are familiar with is also `type`: Python ``` ``` For that matter, the type of `type` is `type` as well (yes, really): Python ``` ``` `type` is a metaclass, of which classes are instances. Just as an ordinary object is an instance of a class, any new-style class in Python, and thus any class in Python 3, is an instance of the `type` metaclass. In the above case: - `x` is an instance of class `Foo`. - `Foo` is an instance of the `type` metaclass. - `type` is also an instance of the `type` metaclass, so it is an instance of itself. [](https://files.realpython.com/media/class-chain.5cb031a299fe.png) [Remove ads](https://realpython.com/account/join/) ## Defining a Class Dynamically The built-in `type()` function, when passed one argument, returns the type of an object. For new-style classes, that is generally the same as the [object’s `__class__` attribute](https://docs.python.org/3/library/stdtypes.html#instance.__class__): Python ``` ``` You can also call `type()` with three arguments—`type(<name>, <bases>, <dct>)`: - `<name>` specifies the class name. This becomes the `__name__` attribute of the class. - `<bases>` specifies a tuple of the base classes from which the class inherits. This becomes the `__bases__` attribute of the class. - `<dct>` specifies a [namespace dictionary](https://realpython.com/python-namespaces-scope/#python-namespace-dictionaries) containing definitions for the class body. This becomes the `__dict__` attribute of the class. Calling `type()` in this manner creates a new instance of the `type` metaclass. In other words, it dynamically creates a new class. In each of the following examples, the top snippet defines a class dynamically with `type()`, while the snippet below it defines the class the usual way, with the `class` statement. In each case, the two snippets are functionally equivalent. ### Example 1 In this first example, the `<bases>` and `<dct>` arguments passed to `type()` are both empty. No [inheritance](https://realpython.com/inheritance-composition-python/) from any parent class is specified, and nothing is initially placed in the namespace dictionary. This is the simplest class definition possible: Python ``` ``` Python ``` ``` ### Example 2 Here, `<bases>` is a tuple with a single element `Foo`, specifying the parent class that `Bar` inherits from. An attribute, `attr`, is initially placed into the namespace dictionary: Python ``` ``` Python ``` ``` ### Example 3 This time, `<bases>` is again empty. Two objects are placed into the namespace dictionary via the `<dct>` argument. The first is an attribute named `attr` and the second a function named `attr_val`, which becomes a method of the defined class: Python ``` ``` Python ``` ``` ### Example 4 Only very simple functions can be defined with [`lambda` in Python](https://dbader.org/blog/python-lambda-functions). In the following example, a slightly more complex function is defined externally then assigned to `attr_val` in the namespace dictionary via the name `f`: Python ``` ``` Python ``` ``` [Remove ads](https://realpython.com/account/join/) ## Custom Metaclasses Consider again this well-worn example: Python ``` ``` The expression `Foo()` creates a new instance of class `Foo`. When the interpreter encounters `Foo()`, the following occurs: - The `__call__()` method of `Foo`’s parent class is called. Since `Foo` is a standard new-style class, its parent class is the `type` metaclass, so `type`’s `__call__()` method is invoked. - That `__call__()` method in turn invokes the following: - `__new__()` - `__init__()` If `Foo` does not define `__new__()` and `__init__()`, default methods are inherited from `Foo`’s ancestry. But if `Foo` does define these methods, they override those from the ancestry, which allows for customized behavior when instantiating `Foo`. In the following, a custom method called `new()` is defined and assigned as the `__new__()` method for `Foo`: Python ``` ``` This modifies the instantiation behavior of class `Foo`: each time an instance of `Foo` is created, by default it is initialized with an attribute called `attr`, which has a value of `100`. (Code like this would more usually appear in the `__init__()` method and not typically in `__new__()`. This example is contrived for demonstration purposes.) Now, as has already been reiterated, classes are objects too. Suppose you wanted to similarly customize instantiation behavior when creating a class like `Foo`. If you were to follow the pattern above, you’d again define a custom method and assign it as the `__new__()` method for the class of which `Foo` is an instance. `Foo` is an instance of the `type` metaclass, so the code looks something like this: Python ``` ``` Except, as you can see, you can’t reassign the `__new__()` method of the `type` metaclass. Python doesn’t allow it. This is probably just as well. `type` is the metaclass from which all new-style classes are derived. You really shouldn’t be mucking around with it anyway. But then what recourse is there, if you want to customize instantiation of a class? One possible solution is a custom metaclass. Essentially, instead of mucking around with the `type` metaclass, you can define your own metaclass, which derives from `type`, and then you can muck around with that instead. The first step is to define a metaclass that derives from `type`, as follows: Python ``` ``` The definition header `class Meta(type):` specifies that `Meta` derives from `type`. Since `type` is a metaclass, that makes `Meta` a metaclass as well. Note that a custom `__new__()` method has been defined for `Meta`. It wasn’t possible to do that to the `type` metaclass directly. The `__new__()` method does the following: - Delegates via `super()` to the `__new__()` method of the parent metaclass (`type`) to actually create a new class - Assigns the custom attribute `attr` to the class, with a value of `100` - Returns the newly created class Now the other half of the voodoo: Define a new class `Foo` and specify that its metaclass is the custom metaclass `Meta`, rather than the standard metaclass `type`. This is done using the `metaclass` keyword in the class definition as follows: Python ``` ``` *Voila\!* `Foo` has picked up the `attr` attribute automatically from the `Meta` metaclass. Of course, any other classes you define similarly will do likewise: Python ``` ``` In the same way that a class functions as a template for the creation of objects, a metaclass functions as a template for the creation of classes. Metaclasses are sometimes referred to as [class factories](https://en.wikipedia.org/wiki/Factory_\(object-oriented_programming\)). Compare the following two examples: **Object Factory:** Python ``` ``` **Class Factory:** Python ``` ``` [Remove ads](https://realpython.com/account/join/) ## Is This Really Necessary? As simple as the above class factory example is, it is the essence of how metaclasses work. They allow customization of class instantiation. Still, this is a lot of fuss just to bestow the custom attribute `attr` on each newly created class. Do you really need a metaclass just for that? In Python, there are at least a couple other ways in which effectively the same thing can be accomplished: **Simple Inheritance:** Python ``` ``` **Class Decorator:** Python ``` ``` ## Conclusion As Tim Peters suggests, **metaclasses** can easily veer into the realm of being a “solution in search of a problem.” It isn’t typically necessary to create custom metaclasses. If the problem at hand can be solved in a simpler way, it probably should be. Still, it is beneficial to understand metaclasses so that you understand [Python classes](https://realpython.com/python3-object-oriented-programming/) in general and can recognize when a metaclass really is the appropriate tool to use. Mark as Completed Share Recommended Course [Metaclasses in Python](https://realpython.com/courses/python-metaclasses/) (51m) 🐍 Python Tricks 💌 Get a short & sweet **Python Trick** delivered to your inbox every couple of days. No spam ever. Unsubscribe any time. Curated by the Real Python team.  About **John Sturtz** [ ](https://realpython.com/team/jsturtz/) John is an avid Pythonista and a member of the Real Python tutorial team. [» More about John](https://realpython.com/team/jsturtz/) *** *Each tutorial at Real Python is created by a team of developers so that it meets our high quality standards. The team members who worked on this tutorial are:* [](https://realpython.com/team/asantos/) [Aldren](https://realpython.com/team/asantos/) [](https://realpython.com/team/dbader/) [Dan](https://realpython.com/team/dbader/) [](https://realpython.com/team/jjablonski/) [Joanna](https://realpython.com/team/jjablonski/) Master Real-World Python Skills With Unlimited Access to Real Python  **Join us and get access to thousands of tutorials, hands-on video courses, and a community of expert Pythonistas:** [Level Up Your Python Skills »](https://realpython.com/account/join/?utm_source=rp_article_footer&utm_content=python-metaclasses) Master Real-World Python Skills With Unlimited Access to Real Python  **Join us and get access to thousands of tutorials, hands-on video courses, and a community of expert Pythonistas:** [Level Up Your Python Skills »](https://realpython.com/account/join/?utm_source=rp_article_footer&utm_content=python-metaclasses) What Do You Think? **Rate this article:** [LinkedIn](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Frealpython.com%2Fpython-metaclasses%2F) [Twitter](https://twitter.com/intent/tweet/?text=Interesting%20Python%20article%20on%20%40realpython%3A%20Python%20Metaclasses&url=https%3A%2F%2Frealpython.com%2Fpython-metaclasses%2F) [Bluesky](https://bsky.app/intent/compose?text=Interesting%20Python%20article%20on%20%40realpython.com%3A%20Python%20Metaclasses%20https%3A%2F%2Frealpython.com%2Fpython-metaclasses%2F) [Facebook](https://facebook.com/sharer/sharer.php?u=https%3A%2F%2Frealpython.com%2Fpython-metaclasses%2F) [Email](mailto:?subject=Python%20article%20for%20you&body=Python%20Metaclasses%20on%20Real%20Python%0A%0Ahttps%3A%2F%2Frealpython.com%2Fpython-metaclasses%2F%0A) What’s your \#1 takeaway or favorite thing you learned? How are you going to put your newfound skills to use? Leave a comment below and let us know. **Commenting Tips:** The most useful comments are those written with the goal of learning from or helping out other students. [Get tips for asking good questions](https://realpython.com/python-beginner-tips/#tip-9-ask-good-questions) and [get answers to common questions in our support portal](https://support.realpython.com/). *** Looking for a real-time conversation? Visit the [Real Python Community Chat](https://realpython.com/community/) or join the next [“Office Hours” Live Q\&A Session](https://realpython.com/office-hours/). Happy Pythoning\! Keep Learning Related Topics: [advanced](https://realpython.com/tutorials/advanced/) [python](https://realpython.com/tutorials/python/) Related Learning Paths: - [Python Metaprogramming](https://realpython.com/learning-paths/metaprogramming-in-python/?utm_source=realpython&utm_medium=web&utm_campaign=related-learning-path&utm_content=python-metaclasses) Related Courses: - [Metaclasses in Python](https://realpython.com/courses/python-metaclasses/?utm_source=realpython&utm_medium=web&utm_campaign=related-course&utm_content=python-metaclasses) Related Tutorials: - [Primer on Python Decorators](https://realpython.com/primer-on-python-decorators/?utm_source=realpython&utm_medium=web&utm_campaign=related-post&utm_content=python-metaclasses) - [Implementing an Interface in Python](https://realpython.com/python-interface/?utm_source=realpython&utm_medium=web&utm_campaign=related-post&utm_content=python-metaclasses) - [Python Descriptors: An Introduction](https://realpython.com/python-descriptors/?utm_source=realpython&utm_medium=web&utm_campaign=related-post&utm_content=python-metaclasses) - [Python's asyncio: A Hands-On Walkthrough](https://realpython.com/async-io-python/?utm_source=realpython&utm_medium=web&utm_campaign=related-post&utm_content=python-metaclasses) - [Iterators and Iterables in Python: Run Efficient Iterations](https://realpython.com/python-iterators-iterables/?utm_source=realpython&utm_medium=web&utm_campaign=related-post&utm_content=python-metaclasses) ## Keep reading Real Python by creating a free account or signing in: [](https://realpython.com/account/signup/?intent=continue_reading&utm_source=rp&utm_medium=web&utm_campaign=rwn&utm_content=v1&next=%2Fpython-metaclasses%2F) [Continue »](https://realpython.com/account/signup/?intent=continue_reading&utm_source=rp&utm_medium=web&utm_campaign=rwn&utm_content=v1&next=%2Fpython-metaclasses%2F) Already have an account? [Sign-In](https://realpython.com/account/login/?next=/python-metaclasses/) Almost there! Complete this form and click the button below to gain instant access: ×  Join the Real Python Community Newsletter (More Than 45,468 Python Developers Have Subscribed) ##### Learn Python - [Start Here](https://realpython.com/start-here/) - [Learning Resources](https://realpython.com/search) - [Code Mentor](https://realpython.com/mentor/) - [Python Reference](https://realpython.com/ref/) - [Python Cheat Sheet](https://realpython.com/cheatsheets/python/) - [Support Center](https://support.realpython.com/) ##### Courses & Paths - [Learning Paths](https://realpython.com/learning-paths/) - [Quizzes & Exercises](https://realpython.com/quizzes/) - [Browse Topics](https://realpython.com/tutorials/all/) - [Live Courses](https://realpython.com/live/) - [Books](https://realpython.com/books/) ##### Community - [Podcast](https://realpython.com/podcasts/rpp/) - [Newsletter](https://realpython.com/newsletter/) - [Community Chat](https://realpython.com/community/) - [Office Hours](https://realpython.com/office-hours/) - [Learner Stories](https://realpython.com/learner-stories/) ##### Membership - [Plans & Pricing](https://realpython.com/account/join/) - [Team Plans](https://realpython.com/account/join-team/) - [For Business](https://realpython.com/account/join-team/inquiry/) - [For Schools](https://realpython.com/account/join-team/education-inquiry/) - [Reviews](https://realpython.com/learner-stories/) ##### Company - [About Us](https://realpython.com/about/) - [Team](https://realpython.com/team/) - [Mission & Values](https://realpython.com/mission/) - [Editorial Guidelines](https://realpython.com/editorial-guidelines/) - [Sponsorships](https://realpython.com/sponsorships/) - [Careers](https://realpython.workable.com/) - [Press Kit](https://realpython.com/media-kit/) - [Merch](https://realpython.com/merch) [Privacy Policy](https://realpython.com/privacy-policy/) ⋅ [Terms of Use](https://realpython.com/terms/) ⋅ [Security](https://realpython.com/security/) ⋅ [Contact](https://realpython.com/contact/) Happy Pythoning\! © 2012–2026 DevCademy Media Inc. DBA Real Python. All rights reserved. REALPYTHON™ is a trademark of DevCademy Media Inc. [](https://realpython.com/)  You've blocked notifications

The term **metaprogramming** refers to the potential for a program to have knowledge of or manipulate itself. Python supports a form of metaprogramming for classes called **metaclasses**. Metaclasses are an esoteric [OOP concept](https://realpython.com/python3-object-oriented-programming/), lurking behind virtually all Python code. You are using them whether you are aware of it or not. For the most part, you don’t need to be aware of it. Most Python programmers rarely, if ever, have to think about metaclasses. When the need arises, however, Python provides a capability that not all object-oriented languages support: you can get under the hood and define custom metaclasses. The use of custom metaclasses is somewhat controversial, as suggested by the following quote from Tim Peters, the Python guru who authored the [Zen of Python](https://realpython.com/zen-of-python/): > “Metaclasses are deeper magic than 99% of users should ever worry about. If you wonder whether you need them, you don’t (the people who actually need them know with certainty that they need them, and don’t need an explanation about why).” > > — *Tim Peters* There are Pythonistas (as Python aficionados are known) who believe that you should never use custom metaclasses. That might be going a bit far, but it is probably true that custom metaclasses mostly aren’t necessary. If it isn’t pretty obvious that a problem calls for them, then it will probably be cleaner and more readable if solved in a simpler way. Still, understanding Python metaclasses is worthwhile, because it leads to a better understanding of the internals of [Python classes](https://realpython.com/python-classes/) in general. You never know: you may one day find yourself in one of those situations where you just know that a custom metaclass is what you want. ## Old-Style vs. New-Style Classes In the Python realm, a class [can be one of two varieties](https://wiki.python.org/moin/NewClassVsClassicClass). No official terminology has been decided on, so they are informally referred to as old-style and new-style classes. ### Old-Style Classes With old-style classes, class and type are not quite the same thing. An instance of an old-style class is always implemented from a single built-in type called `instance`. If `obj` is an instance of an old-style class, `obj.__class__` designates the class, but `type(obj)` is always `instance`. The following example is taken from Python 2.7: ### New-Style Classes New-style classes unify the concepts of class and type. If `obj` is an instance of a new-style class, `type(obj)` is the same as `obj.__class__`: ## Type and Class In Python 3, all classes are new-style classes. Thus, in Python 3 it is reasonable to refer to an object’s type and its class interchangeably. Remember that, [in Python, everything is an object.](https://web.archive.org/web/20151210024637/https://mail.python.org/pipermail/python-list/2015-June/691689.html) Classes are objects as well. As a result, a class must have a type. What is the type of a class? Consider the following: The type of `x` is class `Foo`, as you would expect. But the type of `Foo`, the class itself, is `type`. In general, the type of any new-style class is `type`. The type of the built-in classes you are familiar with is also `type`: For that matter, the type of `type` is `type` as well (yes, really): `type` is a metaclass, of which classes are instances. Just as an ordinary object is an instance of a class, any new-style class in Python, and thus any class in Python 3, is an instance of the `type` metaclass. In the above case: - `x` is an instance of class `Foo`. - `Foo` is an instance of the `type` metaclass. - `type` is also an instance of the `type` metaclass, so it is an instance of itself. [](https://files.realpython.com/media/class-chain.5cb031a299fe.png) ## Defining a Class Dynamically The built-in `type()` function, when passed one argument, returns the type of an object. For new-style classes, that is generally the same as the [object’s `__class__` attribute](https://docs.python.org/3/library/stdtypes.html#instance.__class__): You can also call `type()` with three arguments—`type(<name>, <bases>, <dct>)`: - `<name>` specifies the class name. This becomes the `__name__` attribute of the class. - `<bases>` specifies a tuple of the base classes from which the class inherits. This becomes the `__bases__` attribute of the class. - `<dct>` specifies a [namespace dictionary](https://realpython.com/python-namespaces-scope/#python-namespace-dictionaries) containing definitions for the class body. This becomes the `__dict__` attribute of the class. Calling `type()` in this manner creates a new instance of the `type` metaclass. In other words, it dynamically creates a new class. In each of the following examples, the top snippet defines a class dynamically with `type()`, while the snippet below it defines the class the usual way, with the `class` statement. In each case, the two snippets are functionally equivalent. ### Example 1 In this first example, the `<bases>` and `<dct>` arguments passed to `type()` are both empty. No [inheritance](https://realpython.com/inheritance-composition-python/) from any parent class is specified, and nothing is initially placed in the namespace dictionary. This is the simplest class definition possible: ### Example 2 Here, `<bases>` is a tuple with a single element `Foo`, specifying the parent class that `Bar` inherits from. An attribute, `attr`, is initially placed into the namespace dictionary: ### Example 3 This time, `<bases>` is again empty. Two objects are placed into the namespace dictionary via the `<dct>` argument. The first is an attribute named `attr` and the second a function named `attr_val`, which becomes a method of the defined class: ### Example 4 Only very simple functions can be defined with [`lambda` in Python](https://dbader.org/blog/python-lambda-functions). In the following example, a slightly more complex function is defined externally then assigned to `attr_val` in the namespace dictionary via the name `f`: Consider again this well-worn example: The expression `Foo()` creates a new instance of class `Foo`. When the interpreter encounters `Foo()`, the following occurs: - The `__call__()` method of `Foo`’s parent class is called. Since `Foo` is a standard new-style class, its parent class is the `type` metaclass, so `type`’s `__call__()` method is invoked. - That `__call__()` method in turn invokes the following: - `__new__()` - `__init__()` If `Foo` does not define `__new__()` and `__init__()`, default methods are inherited from `Foo`’s ancestry. But if `Foo` does define these methods, they override those from the ancestry, which allows for customized behavior when instantiating `Foo`. In the following, a custom method called `new()` is defined and assigned as the `__new__()` method for `Foo`: This modifies the instantiation behavior of class `Foo`: each time an instance of `Foo` is created, by default it is initialized with an attribute called `attr`, which has a value of `100`. (Code like this would more usually appear in the `__init__()` method and not typically in `__new__()`. This example is contrived for demonstration purposes.) Now, as has already been reiterated, classes are objects too. Suppose you wanted to similarly customize instantiation behavior when creating a class like `Foo`. If you were to follow the pattern above, you’d again define a custom method and assign it as the `__new__()` method for the class of which `Foo` is an instance. `Foo` is an instance of the `type` metaclass, so the code looks something like this: Except, as you can see, you can’t reassign the `__new__()` method of the `type` metaclass. Python doesn’t allow it. This is probably just as well. `type` is the metaclass from which all new-style classes are derived. You really shouldn’t be mucking around with it anyway. But then what recourse is there, if you want to customize instantiation of a class? One possible solution is a custom metaclass. Essentially, instead of mucking around with the `type` metaclass, you can define your own metaclass, which derives from `type`, and then you can muck around with that instead. The first step is to define a metaclass that derives from `type`, as follows: The definition header `class Meta(type):` specifies that `Meta` derives from `type`. Since `type` is a metaclass, that makes `Meta` a metaclass as well. Note that a custom `__new__()` method has been defined for `Meta`. It wasn’t possible to do that to the `type` metaclass directly. The `__new__()` method does the following: - Delegates via `super()` to the `__new__()` method of the parent metaclass (`type`) to actually create a new class - Assigns the custom attribute `attr` to the class, with a value of `100` - Returns the newly created class Now the other half of the voodoo: Define a new class `Foo` and specify that its metaclass is the custom metaclass `Meta`, rather than the standard metaclass `type`. This is done using the `metaclass` keyword in the class definition as follows: *Voila\!* `Foo` has picked up the `attr` attribute automatically from the `Meta` metaclass. Of course, any other classes you define similarly will do likewise: In the same way that a class functions as a template for the creation of objects, a metaclass functions as a template for the creation of classes. Metaclasses are sometimes referred to as [class factories](https://en.wikipedia.org/wiki/Factory_\(object-oriented_programming\)). Compare the following two examples: **Object Factory:** **Class Factory:** ## Is This Really Necessary? As simple as the above class factory example is, it is the essence of how metaclasses work. They allow customization of class instantiation. Still, this is a lot of fuss just to bestow the custom attribute `attr` on each newly created class. Do you really need a metaclass just for that? In Python, there are at least a couple other ways in which effectively the same thing can be accomplished: **Simple Inheritance:** **Class Decorator:** ## Conclusion As Tim Peters suggests, **metaclasses** can easily veer into the realm of being a “solution in search of a problem.” It isn’t typically necessary to create custom metaclasses. If the problem at hand can be solved in a simpler way, it probably should be. Still, it is beneficial to understand metaclasses so that you understand [Python classes](https://realpython.com/python3-object-oriented-programming/) in general and can recognize when a metaclass really is the appropriate tool to use.