How can implement the equivalent of a __getattr__ on a class, on a module?
Example
When calling a function that does not exist in a module's statically defined attributes, I wish to create an instance of a class in that module, and invoke the method on it with the same name as failed in the attribute lookup on the module.
class A(object):
def salutation(self, accusative):
print "hello", accusative
# note this function is intentionally on the module, and not the class above
def __getattr__(mod, name):
return getattr(A(), name)
if __name__ == "__main__":
# i hope here to have my __getattr__ function above invoked, since
# salutation does not exist in the current namespace
salutation("world")
Which gives:
matt#stanley:~/Desktop$ python getattrmod.py
Traceback (most recent call last):
File "getattrmod.py", line 9, in <module>
salutation("world")
NameError: name 'salutation' is not defined
There are two basic problems you are running into here:
__xxx__ methods are only looked up on the class
TypeError: can't set attributes of built-in/extension type 'module'
(1) means any solution would have to also keep track of which module was being examined, otherwise every module would then have the instance-substitution behavior; and (2) means that (1) isn't even possible... at least not directly.
Fortunately, sys.modules is not picky about what goes there so a wrapper will work, but only for module access (i.e. import somemodule; somemodule.salutation('world'); for same-module access you pretty much have to yank the methods from the substitution class and add them to globals() eiher with a custom method on the class (I like using .export()) or with a generic function (such as those already listed as answers). One thing to keep in mind: if the wrapper is creating a new instance each time, and the globals solution is not, you end up with subtly different behavior. Oh, and you don't get to use both at the same time -- it's one or the other.
Update
From Guido van Rossum:
There is actually a hack that is occasionally used and recommended: a
module can define a class with the desired functionality, and then at
the end, replace itself in sys.modules with an instance of that class
(or with the class, if you insist, but that's generally less useful).
E.g.:
# module foo.py
import sys
class Foo:
def funct1(self, <args>): <code>
def funct2(self, <args>): <code>
sys.modules[__name__] = Foo()
This works because the import machinery is actively enabling this
hack, and as its final step pulls the actual module out of
sys.modules, after loading it. (This is no accident. The hack was
proposed long ago and we decided we liked enough to support it in the
import machinery.)
So the established way to accomplish what you want is to create a single class in your module, and as the last act of the module replace sys.modules[__name__] with an instance of your class -- and now you can play with __getattr__/__setattr__/__getattribute__ as needed.
Note 1: If you use this functionality then anything else in the module, such as globals, other functions, etc., will be lost when the sys.modules assignment is made -- so make sure everything needed is inside the replacement class.
Note 2: To support from module import * you must have __all__ defined in the class; for example:
class Foo:
def funct1(self, <args>): <code>
def funct2(self, <args>): <code>
__all__ = list(set(vars().keys()) - {'__module__', '__qualname__'})
Depending on your Python version, there may be other names to omit from __all__. The set() can be omitted if Python 2 compatibility is not needed.
A while ago, Guido declared that all special method lookups on
new-style classes bypass __getattr__ and __getattribute__. Dunder methods had previously worked on modules - you could, for example, use a module as a context manager simply by defining __enter__ and __exit__, before those tricks broke.
Recently some historical features have made a comeback, the module __getattr__ among them, and so the existing hack (a module replacing itself with a class in sys.modules at import time) should be no longer necessary.
In Python 3.7+, you just use the one obvious way. To customize attribute access on a module, define a __getattr__ function at the module level which should accept one argument (name of attribute), and return the computed value or raise an AttributeError:
# my_module.py
def __getattr__(name: str) -> Any:
...
This will also allow hooks into "from" imports, i.e. you can return dynamically generated objects for statements such as from my_module import whatever.
On a related note, along with the module getattr you may also define a __dir__ function at module level to respond to dir(my_module). See PEP 562 for details.
This is a hack, but you can wrap the module with a class:
class Wrapper(object):
def __init__(self, wrapped):
self.wrapped = wrapped
def __getattr__(self, name):
# Perform custom logic here
try:
return getattr(self.wrapped, name)
except AttributeError:
return 'default' # Some sensible default
sys.modules[__name__] = Wrapper(sys.modules[__name__])
We don't usually do it that way.
What we do is this.
class A(object):
....
# The implicit global instance
a= A()
def salutation( *arg, **kw ):
a.salutation( *arg, **kw )
Why? So that the implicit global instance is visible.
For examples, look at the random module, which creates an implicit global instance to slightly simplify the use cases where you want a "simple" random number generator.
Similar to what #Håvard S proposed, in a case where I needed to implement some magic on a module (like __getattr__), I would define a new class that inherits from types.ModuleType and put that in sys.modules (probably replacing the module where my custom ModuleType was defined).
See the main __init__.py file of Werkzeug for a fairly robust implementation of this.
This is hackish, but...
# Python 2.7
import types
class A(object):
def salutation(self, accusative):
print("hello", accusative)
def farewell(self, greeting, accusative):
print(greeting, accusative)
def AddGlobalAttribute(classname, methodname):
print("Adding " + classname + "." + methodname + "()")
def genericFunction(*args):
return globals()[classname]().__getattribute__(methodname)(*args)
globals()[methodname] = genericFunction
# set up the global namespace
x = 0 # X and Y are here to add them implicitly to globals, so
y = 0 # globals does not change as we iterate over it.
toAdd = []
def isCallableMethod(classname, methodname):
someclass = globals()[classname]()
something = someclass.__getattribute__(methodname)
return callable(something)
for x in globals():
print("Looking at", x)
if isinstance(globals()[x], (types.ClassType, type)):
print("Found Class:", x)
for y in dir(globals()[x]):
if y.find("__") == -1: # hack to ignore default methods
if isCallableMethod(x,y):
if y not in globals(): # don't override existing global names
toAdd.append((x,y))
# Returns:
# ('Looking at', 'A')
# ('Found Class:', 'A')
# ('Looking at', 'toAdd')
# ('Looking at', '__builtins__')
# ('Looking at', 'AddGlobalAttribute')
# ('Looking at', 'register')
# ('Looking at', '__package__')
# ('Looking at', 'salutation')
# ('Looking at', 'farewell')
# ('Looking at', 'types')
# ('Looking at', 'x')
# ('Looking at', 'y')
# ('Looking at', '__name__')
# ('Looking at', 'isCallableMethod')
# ('Looking at', '__doc__')
# ('Looking at', 'codecs')
for x in toAdd:
AddGlobalAttribute(*x)
if __name__ == "__main__":
salutation("world")
farewell("goodbye", "world")
# Returns:
# hello world
# goodbye world
This works by iterating over the all the objects in the global namespace. If the item is a class, it iterates over the class attributes. If the attribute is callable it adds it to the global namespace as a function.
It ignore all attributes which contain "__".
I wouldn't use this in production code, but it should get you started.
Here's my own humble contribution -- a slight embellishment of #Håvard S's highly rated answer, but a bit more explicit (so it might be acceptable to #S.Lott, even though probably not good enough for the OP):
import sys
class A(object):
def salutation(self, accusative):
print "hello", accusative
class Wrapper(object):
def __init__(self, wrapped):
self.wrapped = wrapped
def __getattr__(self, name):
try:
return getattr(self.wrapped, name)
except AttributeError:
return getattr(A(), name)
_globals = sys.modules[__name__] = Wrapper(sys.modules[__name__])
if __name__ == "__main__":
_globals.salutation("world")
Create your module file that has your classes. Import the module. Run getattr on the module you just imported. You can do a dynamic import using __import__ and pull the module from sys.modules.
Here's your module some_module.py:
class Foo(object):
pass
class Bar(object):
pass
And in another module:
import some_module
Foo = getattr(some_module, 'Foo')
Doing this dynamically:
import sys
__import__('some_module')
mod = sys.modules['some_module']
Foo = getattr(mod, 'Foo')
Related
Call me weird if you like, many have before, but I have a large class which I'd like to make extensible with methods loaded from a plugin directory. Essentially, I'm monkey patching the class. What I have almost works but the method loaded doesn't 'see' the globals defined in __main__. Ideally I'd like a way to tell globals() (or whatever mechanism is actually used to locate global variables) to use that existing in __main__. Here is the code I have (trimmed for the sake of brevity):
#!/usr/bin/env python3
import importlib
import os
import types
main_global = "Hi, I'm in main"
class MyClass:
def __init__(self, plugin_dir=None):
if plugin_dir:
self.load_plugins(plugin_dir, ext="plugin")
def load_plugins(self, plugin_dir, ext):
""" Load plugins
Plugins are files in 'plugin_dir' that have the given extension.
The functions defined within are imported as methods of this class.
"""
cls = self.__class__
# First check that we're not importing the same extension twice into
# the same class.
try:
plugins = getattr(cls, "_plugins")
except AttributeError:
plugins = set()
setattr(cls, "_plugins", plugins)
if ext in plugins:
return
plugins.add(ext)
for file in os.listdir(plugin_dir):
if not file.endswith(ext):
continue
filename = os.path.join(plugin_dir, file)
loader = importlib.machinery.SourceFileLoader("bar", filename)
module = types.ModuleType(loader.name)
loader.exec_module(module)
for name in dir(module):
if name.startswith("__"):
continue
obj = getattr(module, name)
if callable(obj):
obj = obj.__get__(self, cls)
setattr(cls, name, obj)
z = MyClass(plugin_dir="plugins")
z.foo("Hello")
And this is 'foo.plugin' from the plugins directory:
#!/usr/bin/env python3
foo_global = "I am global within foo"
def foo(self, value):
print(f"I am foo, called with {self} and {value}")
print(f"foo_global = {foo_global}")
print(f"main_global = {main_global}")
The output is...
I am foo, called with <__main__.MyClass object at 0x7fd4680bfac8> and Hello
foo_global = I am global within foo
Traceback (most recent call last):
File "./plugged", line 55, in <module>
z.foo("Hello")
File "plugins/foo.plugin", line 8, in foo
print(f"main_global = {main_global}")
NameError: name 'main_global' is not defined
I know it all feels a bit 'hacky', but it's become a challenge so please don't flame me on style etc. If there's another way to achieve this aim, I'm all ears.
Thoughts, learned friends?
You can do what you want with a variation of the technique shown in #Martijn Pieters' answer to the the question: How to inject variable into scope with a decorator? tweaked to inject multiple values into a class method.
from functools import wraps
import importlib
import os
from pathlib import Path
import types
main_global = "Hi, I'm in main"
class MyClass:
def __init__(self, plugin_dir=None):
if plugin_dir:
self.load_plugins(plugin_dir, ext="plugin")
def load_plugins(self, plugin_dir, ext):
""" Load plugins
Plugins are files in 'plugin_dir' that have the given extension.
The functions defined within are imported as methods of this class.
"""
cls = self.__class__
# First check that we're not importing the same extension twice into
# the same class.
try:
plugins = getattr(cls, "_plugins")
except AttributeError:
plugins = set()
setattr(cls, "_plugins", plugins)
if ext in plugins:
return
plugins.add(ext)
for file in Path(plugin_dir).glob(f'*.{ext}'):
loader = importlib.machinery.SourceFileLoader("bar", str(file))
module = types.ModuleType(loader.name)
loader.exec_module(module)
namespace = globals()
for name in dir(module):
if name.startswith("__"):
continue
obj = getattr(module, name)
if callable(obj):
obj = inject(obj.__get__(self, cls), namespace)
setattr(cls, name, obj)
def inject(method, namespace):
#wraps(method)
def wrapped(*args, **kwargs):
method_globals = method.__globals__
# Save copies of any of method's global values replaced by the namespace.
replaced = {key: method_globals[key] for key in namespace if key in method_globals}
method_globals.update(namespace)
try:
method(*args[1:], **kwargs)
finally:
method_globals.update(replaced) # Restore any replaced globals.
return wrapped
z = MyClass(plugin_dir="plugins")
z.foo("Hello")
Example output:
I am foo, called with <__main__.MyClass object at 0x0056F670> and Hello
foo_global = I am global within foo
main_global = Hi, I'm in main
You can approach the problem with a factory function and inheritance. Assuming each of your plugins is something like this, defined in a separate importable file:
class MyPlugin:
foo = 'bar'
def extra_method(self):
print(self.foo)
You can use a factory like this:
def MyClassFactory(plugin_dir):
def search_and_import_plugins(plugin_dir):
# Look for all possible plugins and import them
return plugin_list # a list of plugin classes, like [MyPlugin]
plugin_list = search_and_import_plugins(plugin_dir):
class MyClass(*plugin_list):
pass
return MyClass()
z = MyClassFactory('/home/me/plugins')
I'm trying to bypass importing from a module, so in my __init__.py I can inject code like this:
globals().update(
{
"foo": lambda: print("Hello stackoverflow!")
}
)
so if I do import mymodule I will be able to call mymodule.foo. That is a simple concept, useless for the purpose because you can actually just define foo.
So, the idea is to modify the globals module dictionary, so in case it doesn't find the function foo it will go wherever and I can inject the code, for that I tried:
from importer import load #a load function to search for the code
from functools import wraps
def global_get_wrapper(f):
#wraps(f)
def wrapper(*args):
module_name, default = args
res = f(*args)
if res is None:
return load(module_name)
return res
return wrapper
globals().get = global_get_wrapper(globals().get) # trying to substitute get method
But it gives me an error:
AttributeError: 'dict' object attribute 'get' is read-only
The other idea I had is to preload the available function, class, etc names into the module dictionary and lazily load them later.
I run out of ideas to accomplish this and I don't know if this is even possible.
Should I go for writing my own python importer? or is there any other possibility I could not think about?
Thanks in advance.
Instead of hacking globals() it would be better to define __getattr__ for your module as follows:
module_name.py
foo = 'foo'
def bar():
return 'bar'
my_module.py
import sys
import module_name
class MyModule(object):
def foobar(self):
return 'foobar'
def __getattr__(self, item):
return getattr(module_name, item)
sys.modules[__name__] = MyModule()
and then:
>>> import my_module
>>> my_module.foo
'foo'
>>> my_module.bar()
'bar'
>>> my_module.foobar()
'foobar'
PEP 562, which targets Python 3.7, introduces __getattr__ for modules. In the rationale it also describes workarounds for previous Python versions.
It is sometimes convenient to customize or otherwise have control over access to module attributes. A typical example is managing deprecation warnings. Typical workarounds are assigning __class__ of a module object to a custom subclass of types.ModuleType or replacing the sys.modules item with a custom wrapper instance. It would be convenient to simplify this procedure by recognizing __getattr__ defined directly in a module that would act like a normal __getattr__ method, except that it will be defined on module instances.
So your mymodule can look like:
foo = 'bar'
def __getattr__(name):
print('load you custom module and return it')
Here's how it behaves:
>>> import mymodule
>>> mymodule.foo
'bar'
>>> mymodule.baz
load you custom module and return it
I don't quite understand. Would this work for you?
try:
mymodule.foo()
except:
print("whatever you wanted to do")
With this code I want to get "main, file" and "abc", but I get ModelBase, Model etc...
I don't know what I can do. I define abc in b.py so i think it should execute in the global scope of b.py.
Module model.py:
class ModelBase(type):
def __new__(cls,name,bases,attrs):
module = attrs.pop('__module__');
super_new = super(ModelBase,cls).__new__
def test(cls):
# want to print and references globals from Module b
gl = globals()
for i in gl:
print i
new_class = super_new(cls,name,bases,{'__module__':module,'test':test})
return new_class
class Model(object):
__metaclass__ = ModelBase
Module b.py:
from model import *
class abc(Model):
pass
tt = t();
tt.test();
You can find the caller's globals by inspecting the call stack. However, pay attention to the remark in the docs of sys._getframe:
CPython implementation detail: This function should be used for
internal and specialized purposes only. It is not guaranteed to exist
in all implementations of Python.
Anyway, the code you need to get the globals of the immediate caller is
sys._getframe(1).f_globals
Interesting usecase today: I need to migrate a module in our codebase following code changes. The old mynamespace.Document will disappear and I want to ensure smooth migration by replacing this package by a code object that will dynamically import the correct path and migrate the corresponding objects.
In short:
# instanciate a dynamic package, but do not load
# statically submodules
mynamespace.Document = SomeObject()
assert 'submodule' not in mynamespace.Document.__dict__
# and later on, when importing it, the submodule
# is built if not already available in __dict__
from namespace.Document.submodule import klass
c = klass()
A few things to note:
I am not talking only of migrating code. A simple huge sed would in a sense be enough to change the code in order to migrate some imports, and I would not need a dynamic module. I am talking of objects. A website, holding some live/stored objects will need migration. Those objects will be loaded assuming that mynamespace.Document.submodule.klass exists, and that's the reason for the dynamic module. I need to provide the site with something to load.
We cannot, or do not want to change the way objects are unpickled/loaded. For simplicity, let's just say that we want to make sure that the idiom from mynamespace.Document.submodule import klass has to work. I cannot use instead from mynamespace import Document as container; klass = getattr(getattr(container, 'submodule'), 'klass')
What I tried:
import sys
from types import ModuleType
class VerboseModule(ModuleType):
def __init__(self, name, doc=None):
super(VerboseModule, self).__init__(name, doc)
sys.modules[name] = self
def __repr__(self):
return "<%s %s>" % (self.__class__.__name__, self.__name__)
def __getattribute__(self, name):
if name not in ('__name__', '__repr__', '__class__'):
print "fetching attribute %s for %s" % (name, self)
return super(VerboseModule, self).__getattribute__(name)
class DynamicModule(VerboseModule):
"""
This module generates a dummy class when asked for a component
"""
def __getattr__(self, name):
class Dummy(object):
pass
Dummy.__name__ = name
Dummy.__module__ = self
setattr(self, name, Dummy)
return Dummy
class DynamicPackage(VerboseModule):
"""
This package should generate dummy modules
"""
def __getattr__(self, name):
mod = DynamicModule("%s.%s" % (self.__name__, name))
setattr(self, name, mod)
return mod
DynamicModule("foobar")
# (the import prints:)
# fetching attribute __path__ for <DynamicModule foobar>
# fetching attribute DynamicModuleWorks for <DynamicModule foobar>
# fetching attribute DynamicModuleWorks for <DynamicModule foobar>
from foobar import DynamicModuleWorks
print DynamicModuleWorks
DynamicPackage('document')
# fetching attribute __path__ for <DynamicPackage document>
from document.submodule import ButDynamicPackageDoesNotWork
# Traceback (most recent call last):
# File "dynamicmodule.py", line 40, in <module>
# from document.submodule import ButDynamicPackageDoesNotWork
#ImportError: No module named submodule
As you can see the Dynamic Package does not work. I do not understand what is happening because document is not even asked for a ButDynamicPackageDoesNotWork attribute.
Can anyone clarify what is happening; and if/how I can fix this?
The problem is that python will bypass the entry in for document in sys.modules and load the file for submodule directly. Of course this doesn't exist.
demonstration:
>>> import multiprocessing
>>> multiprocessing.heap = None
>>> import multiprocessing.heap
>>> multiprocessing.heap
<module 'multiprocessing.heap' from '/usr/lib/python2.6/multiprocessing/heap.pyc'>
We would expect that heap is still None because python can just pull it out of sys.modules but That doesn't happen. The dotted notation essentially maps directly to {something on python path}/document/submodule.py and an attempt is made to load that directly.
Update
The trick is to override pythons importing system. The following code requires your DynamicModule class.
import sys
class DynamicImporter(object):
"""this class works as both a finder and a loader."""
def __init__(self, lazy_packages):
self.packages = lazy_packages
def load_module(self, fullname):
"""this makes the class a loader. It is given name of a module and expected
to return the module object"""
print "loading {0}".format(fullname)
components = fullname.split('.')
components = ['.'.join(components[:i+1])
for i in range(len(components))]
for component in components:
if component not in sys.modules:
DynamicModule(component)
print "{0} created".format(component)
return sys.modules[fullname]
def find_module(self, fullname, path=None):
"""This makes the class a finder. It is given the name of a module as well as
the package that contains it (if applicable). It is expected to return a
loader for that module if it knows of one or None in which case other methods
will be tried"""
if fullname.split('.')[0] in self.packages:
print "found {0}".format(fullname)
return self
else:
return None
# This is a list of finder objects which is empty by defaule
# It is tried before anything else when a request to import a module is encountered.
sys.meta_path=[DynamicImporter('foo')]
from foo.bar import ThisShouldWork
EDIT: Note that this is a REALLY BAD idea to do in production code. This was just an interesting thing for me. Don't do this at home!
Is it possible to modify __metaclass__ variable for whole program (interpreter) in Python?
This simple example is working:
class ChattyType(type):
def __init__(cls, name, bases, dct):
print "Class init", name
super(ChattyType, cls).__init__(name, bases, dct)
__metaclass__= ChattyType
class Data:
pass
data = Data() # prints "Class init Data"
print data
but I would love to be able change of __metaclass__ to work even in submodules. So for example (file m1.py):
class A:
pass
a=A()
print a
file main.py:
class ChattyType(type):
def __init__(cls, name, bases, dct):
print "Class init", name
super(ChattyType, cls).__init__(name, bases, dct)
__metaclass__= ChattyType
import m1 # and now print "Class init A"
class Data:
pass
data = Data() # print "Class init Data"
print data
I understand that global __metaclass__ is no longer working in Python 3.X, but that is not my concern (my code if proof of concept). So is there any way to accomplish this in Python-2.x?
The "global __metaclass__" feature of Python 2 is designed to work per-module, only (just think what havoc it would wreak, otherwise, by forcing your own metaclass on all library and third-party modules that you imported from that point onwards -- shudder!). If it's very important to you to "secretly" alter the behavior of all modules you're importing from a certain point onwards, for whatever cloak-and-dagger reason, you could play very very dirty tricks with an import hook (at worst by first copying the sources to a temporary location while altering them...) but the effort would be proportionate to the enormity of the deed, which seems appropriate;-)
Okay; IMO this is gross, hairy, dark magic. You shouldn't use it, perhaps ever, but especially not in production code. It is kind of interesting just for curiosity's sake, however.
You can write a custom importer using the mechanisms described in PEP 302, and further discussed in Doug Hellmann's PyMOTW: Modules and Imports. That gives you the tools to accomplish the task you contemplated.
I implemented such an importer, just because I was curious. Essentially, for the modules you specify by means of the class variable __chatty_for__, it will insert a custom type as a __metaclass__ variable in the imported module's __dict__, before the code is evaluated. If the code in question defines its own __metaclass__, that will replace the one pre-inserted by the importer. It would be inadvisable to apply this importer to any modules before carefully considering what it would do to them.
I haven't written many importers, so I may have done one or more silly things while writing this one. If anyone notices flaws / corner cases I missed in the implementation, please leave a comment.
source file 1:
# foo.py
class Foo: pass
source file 2:
# bar.py
class Bar: pass
source file 3:
# baaz.py
class Baaz: pass
and the main event:
# chattyimport.py
import imp
import sys
import types
class ChattyType(type):
def __init__(cls, name, bases, dct):
print "Class init", name
super(ChattyType, cls).__init__(name, bases, dct)
class ChattyImporter(object):
__chatty_for__ = []
def __init__(self, path_entry):
pass
def find_module(self, fullname, path=None):
if fullname not in self.__chatty_for__:
return None
try:
if path is None:
self.find_results = imp.find_module(fullname)
else:
self.find_results = imp.find_module(fullname, path)
except ImportError:
return None
(f,fn,(suf,mode,typ)) = self.find_results
if typ == imp.PY_SOURCE:
return self
return None
def load_module(self, fullname):
#print '%s loading module %s' % (type(self).__name__, fullname)
(f,fn,(suf,mode,typ)) = self.find_results
data = f.read()
if fullname in sys.modules:
module = sys.modules[fullname]
else:
sys.modules[fullname] = module = types.ModuleType(fullname)
module.__metaclass__ = ChattyType
module.__file__ = fn
module.__name__ = fullname
codeobj = compile(data, fn, 'exec')
exec codeobj in module.__dict__
return module
class ChattyImportSomeModules(ChattyImporter):
__chatty_for__ = 'foo bar'.split()
sys.meta_path.append(ChattyImportSomeModules(''))
import foo # prints 'Class init Foo'
import bar # prints 'Class init Bar'
import baaz
Nope. (This is a feature!)