Metaprogramming — Python 3 Patterns, Recipes and Idioms (original) (raw)
- Contributors
- ToDo List
- The remainder are from context, from the book.
- A Note To Readers
- Introduction
- Teaching Support
- Book Development Rules
- Developer Guide
- Part I: Foundations
- Python for Programmers
- Initialization and Cleanup
- Unit Testing & Test-Driven Development
- Python 3 Language Changes
- Decorators
- Metaprogramming
- Basic Metaprogramming
- The Metaclass Hook
* The Metaclass Hook in Python 3 - Example: Self-Registration of Subclasses
* Using Class Decorators
* Using the inspect module - Example: Making a Class “Final”
- Using __init__ vs. __new__ in Metaclasses
- Class Methods and Metamethods
* Intercepting Class Creation
* A Class Decorator Singleton - The __prepare__() Metamethod
- Module-level __metaclass__ Assignment
- Metaclass Conflicts
- Further Reading
- Generators, Iterators, and Itertools
- Comprehensions
- Coroutines, Concurrency & Distributed Systems
- Jython
- Part II: Idioms
- Discovering the Details About Your Platform
- A Canonical Form for Command-Line Programs
- Messenger/Data Transfer Object
- Part III: Patterns
- The Pattern Concept
- The Singleton
- Building Application Frameworks
- Fronting for an Implementation
- StateMachine
- Decorator: Dynamic Type Selection
- Iterators: Decoupling Algorithms from Containers
- Factory: Encapsulating Object Creation
- Function Objects
- Changing the Interface
- Table-Driven Code: Configuration Flexibility
- Observer
- Multiple Dispatching
- Visitor
- Pattern Refactoring
- Projects
Python 3 Patterns, Recipes and Idioms
- Docs »
- Metaprogramming
- Edit on Bitbucket
Note
This chapter is written using Python 2.6 syntax; it will be converted to Python 3 at a later date.
Objects are created by other objects: special objects called “classes” that we can set up to spit out objects that are configured to our liking.
Classes are just objects, and they can be modified the same way:
class Foo: pass ... Foo.field = 42 x = Foo() x.field 42 Foo.field2 = 99 x.field2 99 Foo.method = lambda self: "Hi!" x.method() 'Hi!'
To modify a class, you perform operations on it like any other object. You can add and subtract fields and methods, for example. The difference is that any change you make to a class affects all the objects of that class, even the ones that have already been instantiated.
What creates these special “class” objects? Other special objects, called metaclasses.
The default metaclass is called type
and in the vast majority of cases it does the right thing. In some situations, however, you can gain leverage by modifying the way that classes are produced – typically by performing extra actions or injecting code. When this is the case, you can use metaclass programming to modify the way that some of your class objects are created.
It’s worth re-emphasizing that in the vast majority of cases, you don’t need metaclasses, because it’s a fascinating toy and the temptation to use it everywhere can be overwhelming. Some of the examples in this chapter will show both metaclass and non-metaclass solutions to a problem, so you can see that there’s usually another (often simpler) approach.
Some of the functionality that was previously only available with metaclasses is now available in a simpler form using class decorators. It is still useful, however, to understand metaclasses, and certain results can still be achieved only through metaclass programming.
Basic Metaprogramming¶
So metaclasses create classes, and classes create instances. Normally when we write a class, the default metaclass type
is automatically invoked to create that class, and we aren’t even aware that it’s happening.
It’s possible to explicitly code the metaclass’ creation of a class. type
called with one argument produces the type information of an existing class; type
called with three arguments creates a new class object. The arguments when invoking type
are the name of the class, a list of base classes, and a dictionary giving the namespace for the class (all the fields and methods). So the equivalent of:
is:
Classes are often referred to as “types,” so this reads fairly sensibly: you’re calling a function that creates a new type based on its arguments.
We can also add base classes, fields and methods:
Metaprogramming/MyList.py
def howdy(self, you): print("Howdy, " + you)
MyList = type('MyList', (list,), dict(x=42, howdy=howdy))
ml = MyList() ml.append("Camembert") print(ml) print(ml.x) ml.howdy("John")
print(ml.class.class)
""" Output: ['Camembert'] 42 Howdy, John """
Note that printing the class of the class produces the metaclass.
The ability to generate classes programmatically using type
opens up some interesting possibilities. Consider the GreenHouseLanguage.py example in the Jython chapter – all the subclasses in that case were written using repetetive code. We can automate the generation of the subclasses using type
:
Metaprogramming/GreenHouse.py
class Event(object): events = [] # static
def __init__(self, action, time):
self.action = action
self.time = time
Event.events.append(self)
def __cmp__ (self, other):
"So sort() will compare only on time."
return cmp(self.time, other.time)
def run(self):
print("%.2f: %s" % (self.time, self.action))
@staticmethod
def run_events():
Event.events.sort();
for e in Event.events:
e.run()
def create_mc(description):
"Create subclass using the 'type' metaclass"
class_name = "".join(x.capitalize() for x in description.split())
def init(self, time):
Event.init(self, description + " [mc]", time)
globals()[class_name] =
type(class_name, (Event,), dict(init = init))
def create_exec(description): "Create subclass by exec-ing a string" class_name = "".join(x.capitalize() for x in description.split()) klass = """ class %s(Event): def init(self, time): Event.init(self, "%s [exec]", time) """ % (class_name, description) exec klass in globals()
if name == "main":
descriptions = ["Light on", "Light off", "Water on", "Water off",
"Thermostat night", "Thermostat day", "Ring bell"]
initializations = "ThermostatNight(5.00); LightOff(2.00);
WaterOn(3.30); WaterOff(4.45); LightOn(1.00);
RingBell(7.00); ThermostatDay(6.00)"
[create_mc(dsc) for dsc in descriptions]
exec initializations in globals()
[create_exec(dsc) for dsc in descriptions]
exec initializations in globals()
Event.run_events()
""" Output: 1.00: Light on [mc] 1.00: Light on [exec] 2.00: Light off [mc] 2.00: Light off [exec] 3.30: Water on [mc] 3.30: Water on [exec] 4.45: Water off [mc] 4.45: Water off [exec] 5.00: Thermostat night [mc] 5.00: Thermostat night [exec] 6.00: Thermostat day [mc] 6.00: Thermostat day [exec] 7.00: Ring bell [mc] 7.00: Ring bell [exec] """
The Event
base class is the same. The classes are created automatically using the create_mc()
function, which takes itsdescription
argument and generates a class name from it. Then it defines an __init__()
method, which it puts into the namespace dictionary for the type
call, producing a new subclass ofEvent
. Note that the resulting class must be inserted into the global namespace, otherwise it will not be seen.
This approach works fine, but then consider the subsequentcreate_exec()
function, which accomplishes the same thing by calling exec
on a string defining the class. This will be much easier to understand by the vast majority of the people reading your code: those who do not understand metaclasses.
The Metaclass Hook¶
So far, we’ve only used the type
metaclass directly. Metaclass programming involves hooking our own operations into the creation of class objects. This is accomplished by:
- Writing a subclass of the metaclass
type
.- Inserting the new metaclass into the class creation process using the metaclass hook.
In Python 2.x, the metaclass hook is a static field in the class called __metaclass__
. In the ordinary case, this is not assigned so Python just uses type
to create the class. But if you define__metaclass__
to point to a callable, Python will call__metaclass__()
after the initial creation of the class object, passing in the class object, the class name, the list of base classes and the namespace dictionary.
Python 2.x also allows you to assign to the global __metaclass__
hook, which will be used if there is not a class-local__metaclass__
hook (is there an equivalent in Python 3?).
Thus, the basic process of metaclass programming looks like this:
Metaprogramming/SimpleMeta1.py
Two-step metaclass creation in Python 2.x
class SimpleMeta1(type): def init(cls, name, bases, nmspc): super(SimpleMeta1, cls).init(name, bases, nmspc) cls.uses_metaclass = lambda self : "Yes!"
class Simple1(object): metaclass = SimpleMeta1 def foo(self): pass @staticmethod def bar(): pass
simple = Simple1() print([m for m in dir(simple) if not m.startswith('__')])
A new method has been injected by the metaclass:
print simple.uses_metaclass()
""" Output: ['bar', 'foo', 'uses_metaclass'] Yes! """
By convention, when defining metaclasses cls
is used rather thanself
as the first argument to all methods except __new__()
(which uses mcl
, for reasons explained later). cls
is the class object that is being modified.
Note that the practice of calling the base-class constructor first (viasuper()
) in the derived-class constructor should be followed with metaclasses as well.
__metaclass__
only needs to be callable, so in Python 2.x it’s possible to define __metaclass__
inline:
Metaprogramming/SimpleMeta2.py
Combining the steps for metaclass creation in Python 2.x
class Simple2(object): class metaclass(type): def init(cls, name, bases, nmspc): # This won't work: # super(metaclass, cls).init(name, bases, nmspc) # Less-flexible specific call: type.init(cls, name, bases, nmspc) cls.uses_metaclass = lambda self : "Yes!"
class Simple3(Simple2): pass simple = Simple3() print simple.uses_metaclass()
""" Output: Yes! """
The compiler won’t accept the super()
call because it says__metaclass__
hasn’t been defined, forcing us to use the specific call to type.__init__()
.
Because it only needs to be callable, it’s even possible to define__metaclass__
as a function:
Metaprogramming/SimpleMeta3.py
A function for metaclass in Python 2.x
class Simple4(object): def metaclass(name, bases, nmspc): cls = type(name, bases, nmspc) cls.uses_metaclass = lambda self : "Yes!" return cls
simple = Simple4() print simple.uses_metaclass()
""" Output: Yes! """
As you’ll see, Python 3 doesn’t allow the syntax of these last two examples. Even so, the above example makes it quite clear what’s happening: the class object is created, then modified, then returned.
Note
Or does it allow that syntax?
The Metaclass Hook in Python 3¶
Python 3 changes the metaclass hook. It doesn’t disallow the__metaclass__
field, but it ignores it. Instead, you use a keyword argument in the base-class list:
class Simple1(object, metaclass = SimpleMeta1): ...
This means that none of the (clever) alternative ways of defining__metaclass__
directly as a class or function are available in Python 3 [[check this]]. All metaclasses must be defined as separate classes. This is probably just as well, as it makes metaclass programs more consistent and thus easier to read and understand.
Example: Self-Registration of Subclasses¶
It is sometimes convienient to use inheritance as an organizing mechanism – each sublclass becomes an element of a group that you work on. For example, in CodeManager.py in the Comprehensionschapter, the subclasses of Language were all the languages that needed to be processed. Each Language subclass described specific processing traits for that language.
To solve this problem, consider a system that automatically keeps a list of all of its “leaf” subclasses (only the classes that have no inheritors). This way we can easily enumerate through all the subtypes:
Metaprogramming/RegisterLeafClasses.py
class RegisterLeafClasses(type): def init(cls, name, bases, nmspc): super(RegisterLeafClasses, cls).init(name, bases, nmspc) if not hasattr(cls, 'registry'): cls.registry = set() cls.registry.add(cls) cls.registry -= set(bases) # Remove base classes # Metamethods, called on class objects: def iter(cls): return iter(cls.registry) def str(cls): if cls in cls.registry: return cls.name return cls.name + ": " + ", ".join([sc.name for sc in cls])
class Color(object): metaclass = RegisterLeafClasses
class Blue(Color): pass class Red(Color): pass class Green(Color): pass class Yellow(Color): pass print(Color) class PhthaloBlue(Blue): pass class CeruleanBlue(Blue): pass print(Color) for c in Color: # Iterate over subclasses print(c)
class Shape(object): metaclass = RegisterLeafClasses
class Round(Shape): pass class Square(Shape): pass class Triangular(Shape): pass class Boxy(Shape): pass print(Shape) class Circle(Round): pass class Ellipse(Round): pass print(Shape)
""" Output: Color: Red, Blue, Green, Yellow Color: Red, CeruleanBlue, Green, PhthaloBlue, Yellow Red CeruleanBlue Green PhthaloBlue Yellow Shape: Square, Round, Boxy, Triangular Shape: Square, Ellipse, Circle, Boxy, Triangular """
Two separate tests are used to show that the registries are independent of each other. Each test shows what happens when another level of leaf classes are added – the former leaf becomes a base class, and so is removed from the registry.
This also introduces metamethods, which are defined in the metaclass so that they become methods of the class. That is, you call them on the class rather than object instances, and their first argument is the class object rather than self
.
Using Class Decorators¶
Using the inspect module¶
(As in the Comprehensions chapter)
Example: Making a Class “Final”¶
It is sometimes convenient to prevent a class from being inherited:
Metaprogramming/Final.py
Emulating Java's 'final'
class final(type): def init(cls, name, bases, namespace): super(final, cls).init(name, bases, namespace) for klass in bases: if isinstance(klass, final): raise TypeError(str(klass.name) + " is final")
class A(object): pass
class B(A): metaclass= final
print B.bases print isinstance(B, final)
Produces compile-time error:
class C(B): pass
""" Output: (<class '__main__.A'>,) True ... TypeError: B is final """
During class object creation, we check to see if any of the bases are derived from final
. Notice that using a metaclass makes the new type an instance of that metaclass, even though the metaclass doesn’t show up in the base-class list.
Because this process of checking for finality must be installed to happen as the subclasses are created, rather than afterwards as performed by class decorators, it appears that this is an example of something that requires metaclasses and can’t be accomplished with class decorators.
Using __init__
vs. __new__
in Metaclasses¶
It can be confusing when you see metaclass examples that appear to arbitrarily use __new__
or __init__
– why choose one over the other?
__new__
is called for the creation of a new class, while__init__
is called after the class is created, to perform additional initialization before the class is handed to the caller:
Metaprogramming/NewVSInit.py
from pprint import pprint
class Tag1: pass class Tag2: pass class Tag3: def tag3_method(self): pass
class MetaBase(type): def new(mcl, name, bases, nmspc): print('MetaBase.__new__\n') return super(MetaBase, mcl).new(mcl, name, bases, nmspc)
def __init__(cls, name, bases, nmspc):
print('MetaBase.__init__\n')
super(MetaBase, cls).__init__(name, bases, nmspc)
class MetaNewVSInit(MetaBase):
def new(mcl, name, bases, nmspc):
# First argument is the metaclass MetaNewVSInit
print('MetaNewVSInit.new')
for x in (mcl, name, bases, nmspc): pprint(x)
print('')
# These all work because the class hasn't been created yet:
if 'foo' in nmspc: nmspc.pop('foo')
name += '_x'
bases += (Tag1,)
nmspc['baz'] = 42
return super(MetaNewVSInit, mcl).new(mcl, name, bases, nmspc)
def __init__(cls, name, bases, nmspc):
# First argument is the class being initialized
print('MetaNewVSInit.__init__')
for x in (cls, name, bases, nmspc): pprint(x)
print('')
if 'bar' in nmspc: nmspc.pop('bar') # No effect
name += '_y' # No effect
bases += (Tag2,) # No effect
nmspc['pi'] = 3.14159 # No effect
super(MetaNewVSInit, cls).__init__(name, bases, nmspc)
# These do work because they operate on the class object:
cls.__name__ += '_z'
cls.__bases__ += (Tag3,)
cls.e = 2.718
class Test(object): metaclass = MetaNewVSInit def init(self): print('Test.init') def foo(self): print('foo still here') def bar(self): print('bar still here')
t = Test() print('class name: ' + Test.name) print('base classes: ', [c.name for c in Test.bases]) print([m for m in dir(t) if not m.startswith("__")]) t.bar() print(t.e)
""" Output: MetaNewVSInit.new <class '__main__.MetaNewVSInit'> 'Test' (<type 'object'>,) {'init': <function __init__ at 0x7ecf0>, 'metaclass': <class '__main__.MetaNewVSInit'>, 'module': 'main', 'bar': <function bar at 0x7ed70>, 'foo': <function foo at 0x7ed30>}
MetaBase.new
MetaNewVSInit.init <class '__main__.Test_x'> 'Test' (<type 'object'>,) {'init': <function __init__ at 0x7ecf0>, 'metaclass': <class '__main__.MetaNewVSInit'>, 'module': 'main', 'bar': <function bar at 0x7ed70>, 'baz': 42}
MetaBase.init
Test.init class name: Test_x_z ('base classes: ', ['object', 'Tag1', 'Tag3']) ['bar', 'baz', 'e', 'tag3_method'] bar still here 2.718 """
The primary difference is that when overriding __new__()
you can change things like the ‘name’, ‘bases’ and ‘namespace’ arguments before you call the super constructor and it will have an effect, but doing the same thing in __init__()
you won’t get any results from the constructor call.
One special case in __new__()
is that you can manipulate things like __slots__
, but in __init__()
you can’t.
Note that, since the base-class version of __init__()
doesn’t make any modifications, it makes sense to call it first, then perform any additional operations. In C++ and Java, the base-class constructor_must_ be called as the first operation in a derived-class constructor, which makes sense because derived-class constructions can then build upon base-class foundations.
In many cases, the choice of __new__()
vs __init__()
is a style issue and doesn’t matter, but because __new__()
can do everything and __init__()
is slightly more limited, some people just start using __new__()
and stick with it. This use can be confusing – I tend to hunt for the reason that__init__()
has been chosen, and if I can’t find it wonder whether the author knew what they were doing. I prefer to only use __new__()
when it has meaning – when you must in order to change things that only __new__()
can change.
Class Methods and Metamethods¶
A metamethod can be called from either the metaclass or from the class, but not from an instance. A classmethod can be called from either a class or its instances, but is not part of the metaclass.
(Is a similar relationship true with attributes, or is it different?)
Intercepting Class Creation¶
This example implements Singleton using metaclasses, by overriding the__call__()
metamethod, which is invoked when a new instance is created:
Metaprogramming/Singleton.py
class Singleton(type): instance = None def call(cls, *args, **kw): if not cls.instance: cls.instance = super(Singleton, cls).call(*args, **kw) return cls.instance
class ASingleton(object): metaclass = Singleton
a = ASingleton() b = ASingleton() assert a is b print(a.class.name, b.class.name)
class BSingleton(object): metaclass = Singleton
c = BSingleton() d = BSingleton() assert c is d print(c.class.name, d.class.name) assert c is not a
""" Output: ('ASingleton', 'ASingleton') ('BSingleton', 'BSingleton') """
By overriding __call__()
in the metaclass, the creation of instances are intercepted. Instance creation is bypassed if one already exists.
Note the dependence upon the behavior of static class fields. Whencls.instance
is first read, it gets the static value ofinstance
from the metaclass, which is None
. However, when the assignment is made, Python creates a local version for the particular class, and the next time cls.instance
is read, it sees that local version. Because of this behavior, each class ends up with its own class-specific instance
field (thus instance
is not somehow being “inherited” from the metaclass).
A Class Decorator Singleton¶
Metaprogramming/SingletonDecorator.py
def singleton(klass): "Simple replacement of object creation operation" def getinstance(*args, **kw): if not hasattr(klass, 'instance'): klass.instance = klass(*args, **kw) return klass.instance return getinstance
def singleton(klass): """ More powerful approach: Change the behavior of the instances AND the class object. """ class Decorated(klass): def init(self, *args, **kwargs): if hasattr(klass, 'init'): klass.init(self, *args, **kwargs) def repr(self) : return klass.name + " obj" str = repr Decorated.name = klass.name class ClassObject: def init(cls): cls.instance = None def repr(cls): return klass.name str = repr def call(cls, *args, **kwargs): print str(cls) + " call " if not cls.instance: cls.instance = Decorated(*args, **kwargs) return cls.instance return ClassObject()
@singleton class ASingleton: pass
a = ASingleton() b = ASingleton() print(a, b) print a.class.name print ASingleton assert a is b
@singleton class BSingleton: def init(self, x): self.x = x
c = BSingleton(11) d = BSingleton(22) assert c is d assert c is not a
""" Output: ASingleton call ASingleton call (ASingleton obj, ASingleton obj) ASingleton ASingleton BSingleton call BSingleton call """
The __prepare__()
Metamethod¶
One of the things you can’t do with class decorators is to replace the default dictionary. In Python 3 this is enabled with the__prepare__()
metamethod:
@classmethod def prepare(mcl, name, bases): return odict()
For an example of using both __prepare__()
and __slots__
in metaclasses, see Michele Simionato’s article.
Module-level __metaclass__
Assignment¶
(Does this work in Python 3? If not is there an alternative?)
Metaclass Conflicts¶
Note that the metaclass
argument is singular – you can’t attach more than one metaclass to a class. However, through multiple inheritance you can accidentally end up with more than one metaclass, and this produces a conflict which must be resolved.
http://code.activestate.com/recipes/204197/