In my python package I have an entry_point run.py file which takes the seed (e.g. 42) and the cuda device (e.g. "cuda:0") as command line argument.
Since both of these variables are used throughout the entire package at different places, I don't want to pass them as arguments from function to function. Hence, I did the following:
utils.py:
import random
import numpy as np
import torch
def set_device(device: str):
global _DEVICE
_DEVICE = torch.device(device)
def get_device() -> torch.device:
return _DEVICE
def set_seed_number(seed: int):
global _SEED
_SEED = seed
def set_seeds():
torch.manual_seed(_SEED)
random.seed(_SEED)
np.random.seed(_SEED)
And then within run.py I set these variables once by calling:
from package.utils import set_device, set_seed_number
...
set_device(device)
set_seed_number(seed=seed)
Now I can import and call the get_device()and set_seeds method from anywhere in my package and I don't have to pass these variables as arguments.
So far this approach works fine, but after reading that using globals in python is strongly discouraged I am wondering if there is a more pythonic way of achieving the above discribed goal?
I already thought of having a dedicated Singleton class, which dynamically would instantiate those constants but I am not exactly sure if and how that would work and if it would be considered more "pythonic" after all.
Thanks already for your answers and maybe you can point me to some patterns that seem applicable in this situation. I can only guess that I am not the first one trying to achieve the above discribed goal.
I can't honestly see a problem with global if it is used sparingly and only when there is a strong reason to do so. (I think the strong discouragement aganst global is because it is often abused.)
But as regards your proposed custom class, there is no need to instantiate it -- you can just set class variables.
main.py
import settings
settings.set_foo(3)
print(settings.Settings.foo)
settings.py
class Settings:
pass
def set_foo(x):
Settings.foo = x
This is no different in principle from putting your data items inside some other mutable collection e.g. a dictionary and then setting them inside functions in the module that defines it (or another that imports it).
main.py
import settings
settings.set_foo(3)
print(settings.settings['foo'])
settings.py
settings = {}
def set_foo(x):
settings['foo'] = x
Related
I've run into a bit of a wall importing modules in a Python script. I'll do my best to describe the error, why I run into it, and why I'm tying this particular approach to solve my problem (which I will describe in a second):
Let's suppose I have a module in which I've defined some utility functions/classes, which refer to entities defined in the namespace into which this auxiliary module will be imported (let "a" be such an entity):
module1:
def f():
print a
And then I have the main program, where "a" is defined, into which I want to import those utilities:
import module1
a=3
module1.f()
Executing the program will trigger the following error:
Traceback (most recent call last):
File "Z:\Python\main.py", line 10, in <module>
module1.f()
File "Z:\Python\module1.py", line 3, in f
print a
NameError: global name 'a' is not defined
Similar questions have been asked in the past (two days ago, d'uh) and several solutions have been suggested, however I don't really think these fit my requirements. Here's my particular context:
I'm trying to make a Python program which connects to a MySQL database server and displays/modifies data with a GUI. For cleanliness sake, I've defined the bunch of auxiliary/utility MySQL-related functions in a separate file. However they all have a common variable, which I had originally defined inside the utilities module, and which is the cursor object from MySQLdb module.
I later realised that the cursor object (which is used to communicate with the db server) should be defined in the main module, so that both the main module and anything that is imported into it can access that object.
End result would be something like this:
utilities_module.py:
def utility_1(args):
code which references a variable named "cur"
def utility_n(args):
etcetera
And my main module:
program.py:
import MySQLdb, Tkinter
db=MySQLdb.connect(#blahblah) ; cur=db.cursor() #cur is defined!
from utilities_module import *
And then, as soon as I try to call any of the utilities functions, it triggers the aforementioned "global name not defined" error.
A particular suggestion was to have a "from program import cur" statement in the utilities file, such as this:
utilities_module.py:
from program import cur
#rest of function definitions
program.py:
import Tkinter, MySQLdb
db=MySQLdb.connect(#blahblah) ; cur=db.cursor() #cur is defined!
from utilities_module import *
But that's cyclic import or something like that and, bottom line, it crashes too. So my question is:
How in hell can I make the "cur" object, defined in the main module, visible to those auxiliary functions which are imported into it?
Thanks for your time and my deepest apologies if the solution has been posted elsewhere. I just can't find the answer myself and I've got no more tricks in my book.
Globals in Python are global to a module, not across all modules. (Many people are confused by this, because in, say, C, a global is the same across all implementation files unless you explicitly make it static.)
There are different ways to solve this, depending on your actual use case.
Before even going down this path, ask yourself whether this really needs to be global. Maybe you really want a class, with f as an instance method, rather than just a free function? Then you could do something like this:
import module1
thingy1 = module1.Thingy(a=3)
thingy1.f()
If you really do want a global, but it's just there to be used by module1, set it in that module.
import module1
module1.a=3
module1.f()
On the other hand, if a is shared by a whole lot of modules, put it somewhere else, and have everyone import it:
import shared_stuff
import module1
shared_stuff.a = 3
module1.f()
… and, in module1.py:
import shared_stuff
def f():
print shared_stuff.a
Don't use a from import unless the variable is intended to be a constant. from shared_stuff import a would create a new a variable initialized to whatever shared_stuff.a referred to at the time of the import, and this new a variable would not be affected by assignments to shared_stuff.a.
Or, in the rare case that you really do need it to be truly global everywhere, like a builtin, add it to the builtin module. The exact details differ between Python 2.x and 3.x. In 3.x, it works like this:
import builtins
import module1
builtins.a = 3
module1.f()
As a workaround, you could consider setting environment variables in the outer layer, like this.
main.py:
import os
os.environ['MYVAL'] = str(myintvariable)
mymodule.py:
import os
myval = None
if 'MYVAL' in os.environ:
myval = os.environ['MYVAL']
As an extra precaution, handle the case when MYVAL is not defined inside the module.
This post is just an observation for Python behaviour I encountered. Maybe the advices you read above don't work for you if you made the same thing I did below.
Namely, I have a module which contains global/shared variables (as suggested above):
#sharedstuff.py
globaltimes_randomnode=[]
globalist_randomnode=[]
Then I had the main module which imports the shared stuff with:
import sharedstuff as shared
and some other modules that actually populated these arrays. These are called by the main module. When exiting these other modules I can clearly see that the arrays are populated. But when reading them back in the main module, they were empty. This was rather strange for me (well, I am new to Python). However, when I change the way I import the sharedstuff.py in the main module to:
from globals import *
it worked (the arrays were populated).
Just sayin'
A function uses the globals of the module it's defined in. Instead of setting a = 3, for example, you should be setting module1.a = 3. So, if you want cur available as a global in utilities_module, set utilities_module.cur.
A better solution: don't use globals. Pass the variables you need into the functions that need it, or create a class to bundle all the data together, and pass it when initializing the instance.
The easiest solution to this particular problem would have been to add another function within the module that would have stored the cursor in a variable global to the module. Then all the other functions could use it as well.
module1:
cursor = None
def setCursor(cur):
global cursor
cursor = cur
def method(some, args):
global cursor
do_stuff(cursor, some, args)
main program:
import module1
cursor = get_a_cursor()
module1.setCursor(cursor)
module1.method()
Since globals are module specific, you can add the following function to all imported modules, and then use it to:
Add singular variables (in dictionary format) as globals for those
Transfer your main module globals to it
.
addglobals = lambda x: globals().update(x)
Then all you need to pass on current globals is:
import module
module.addglobals(globals())
Since I haven't seen it in the answers above, I thought I would add my simple workaround, which is just to add a global_dict argument to the function requiring the calling module's globals, and then pass the dict into the function when calling; e.g:
# external_module
def imported_function(global_dict=None):
print(global_dict["a"])
# calling_module
a = 12
from external_module import imported_function
imported_function(global_dict=globals())
>>> 12
The OOP way of doing this would be to make your module a class instead of a set of unbound methods. Then you could use __init__ or a setter method to set the variables from the caller for use in the module methods.
Update
To test the theory, I created a module and put it on pypi. It all worked perfectly.
pip install superglobals
Short answer
This works fine in Python 2 or 3:
import inspect
def superglobals():
_globals = dict(inspect.getmembers(
inspect.stack()[len(inspect.stack()) - 1][0]))["f_globals"]
return _globals
save as superglobals.py and employ in another module thusly:
from superglobals import *
superglobals()['var'] = value
Extended Answer
You can add some extra functions to make things more attractive.
def superglobals():
_globals = dict(inspect.getmembers(
inspect.stack()[len(inspect.stack()) - 1][0]))["f_globals"]
return _globals
def getglobal(key, default=None):
"""
getglobal(key[, default]) -> value
Return the value for key if key is in the global dictionary, else default.
"""
_globals = dict(inspect.getmembers(
inspect.stack()[len(inspect.stack()) - 1][0]))["f_globals"]
return _globals.get(key, default)
def setglobal(key, value):
_globals = superglobals()
_globals[key] = value
def defaultglobal(key, value):
"""
defaultglobal(key, value)
Set the value of global variable `key` if it is not otherwise st
"""
_globals = superglobals()
if key not in _globals:
_globals[key] = value
Then use thusly:
from superglobals import *
setglobal('test', 123)
defaultglobal('test', 456)
assert(getglobal('test') == 123)
Justification
The "python purity league" answers that litter this question are perfectly correct, but in some environments (such as IDAPython) which is basically single threaded with a large globally instantiated API, it just doesn't matter as much.
It's still bad form and a bad practice to encourage, but sometimes it's just easier. Especially when the code you are writing isn't going to have a very long life.
I know there are ways to perform dynamic import of Python modules themselves, but I would like to know if there's a way to write a module such that it can dynamically create its own module contents on demand. I am imagining a module hook that looks something like:
# In some_module.py:
def __import_name__(name):
return some_object
Such that if I were to write from some_module import foo in a script, Python will call some_module.__import_name__("foo") and let me dynamically create and return the contents.
I haven't found anything that works like this exactly in the documentation, though there are references to an "import protocol" with "finders" and "loaders" and "meta hooks" and "import path hooks" that permit customization of the import logic, and I imagine that such a thing is possible.
I discovered you can modify the behavior of a Module from within itself in arbitrary ways by setting sys.modules[__name__].__class__ to a class that implements whatever your chosen behavior.
import sys
import types
class DynamicModule(types.ModuleType):
# This function is what gets called on `from this_module import whatever`
# or `this_module.whatever` accesses.
def __getattr__(self, name):
# This check ensures we don't intercept special values like __path__
# if they're not set elsewhere.
if name.startswith("__") and name.endswith("__"):
return self.__getattribute__(name)
return make_object(name)
# Helpful to define this here if you need to dynamically construct the
# full set of available attributes.
#property
def __all__(self):
return get_all_objects()
# This ensures the DynamicModule class is used to define the behavior of
# this module.
sys.modules[__name__].__class__ = DynamicModule
Something about this feels like it may not be the intended path to do something like this, though, and that I should be hooking into the importlib machinery.
This question already has answers here:
Short description of the scoping rules?
(9 answers)
Closed 1 year ago.
Context: I'm writing a translator from one Python API to another, both in Python 3.5+. I load the file to be translated with a class named FileLoader, described by Fileloader.py. This file loader allows me to transfer the file's content to other classes doing the translation job.
All of the .py files describing each class are in the same folder
I tried two different ways to import my FileLoader module inside the other modules containing the classes doing the translation job. One seems to work, but the other didn't and I don't understand why.
Here are two code examples illustrating both ways:
The working way
import FileLoader
class Parser:
#
def __init__(self, fileLoader):
if isinstance(fileLoader, FileLoader.FileLoader)
self._fileLoader = fileLoader
else:
# raise a nice exception
The crashing way
class Parser:
import FileLoader
#
def __init__(self, fileLoader):
if isinstance(fileLoader, FileLoader.FileLoader)
self._fileLoader = fileLoader
else:
# raise a nice exception
I thought doing the import inside the class's scope (where it's the only scope FileLoader is used) would be enough, since it would know how to relate to the FileLoader module and its content. I'm obviously wrong since it's the first way which worked.
What am I missing about scopes in Python? Or is it about something different?
2 things : this won't work. And there is no benefit to doing it this way.
First, why not?
class Parser:
#this assigns to the Parser namespace, to refer to it
#within a method you need to use `self.FileLoader` or
#Parser.FileLoader
import FileLoader
#`FileLoader` works fine here, under the Parser indentation
#(in its namespace, but outside of method)
copy_of_FileLoader = FileLoader
#
def __init__(self, fileLoader):
# you need to refer to modules under in Parser namespace
# with that `self`, just like you would with any other
# class or instance variable 👇
if isinstance(fileLoader, self.FileLoader.FileLoader)
self._fileLoader = fileLoader
else:
# raise a nice exception
#works here again, since we are outside of method,
#in `Parser` scope/indent.
copy2_of_FileLoader = FileLoader
Second it's not Pythonic and it doesn't help
Customary for the Python community would be to put import FileLoader at the top of the program. Since it seems to be one of your own modules, it would go after std library imports and after third party module imports. You would not put it under a class declaration.
Unless... you had a good (probably bad actually reason to).
My own code, and this doesn't reflect all that well on me, sometimes has stuff like.
class MainManager(batchhelper.BatchManager):
....
def _load(self, *args, **kwargs):
👉 from pssystem.models import NotificationConfig
So, after stating this wasn't a good thing, why am I doing this?
Well, there are some specific circumstances to my code going here. This is a batch, command-line, script, usable within a Django context and it uses some Django ORM models. In order for those to be used, Django needs to be imported first and then setup. But that often happens too early in the context of these types of batch programs and I get circular import errors, with Django complaining that it hasn't initialized yet.
The solution? Defer execution until the method is called, when all the other modules have been imported and Django has been setup elsewhere.
NotificationConfig is now available, but only within that method as it is a local variable in it. It works, but... it's really not great practice.
Remember: anything in the global scope gets executed at module load time, anything under classes at module load time, anything withing method/function bodies when the method/function is called.
#happens at module load time, you could have circular import errors
import X1
class DoImportsLater:
.
#happens at module load time, you could have circular import errors
import X2
def _load(self, *args, **kwargs):
#only happens when this method is called, if ever
#so you shouldn't be seeing circular imports
import X3
import X1 is std practice, Pythonic.
import X2, what are doing, is not and doesn't help
import X3, what I did, is a hack and is covering up circular import references. But it "fixes" the issue.
I am making a tiny framework for games with pygame, on which I wish to implement basic code to quickly start new projects. This will be a module that whoever uses should just create a folder with subfolders for sprite classes, maps, levels, etc.
My question is, how should my framework module load these client modules? I was considering to design it so the developer could just pass to the main object the names of the directories, like:
game = Game()
game.scenarios = 'scenarios'
Then game will append 'scenarios' to sys.path and use __import__(). I've tested and it works.
But then I researched a little more to see if there were already some autoloader in python, so I could avoid to rewrite it, and I found this question Python modules autoloader?
Basically, it is not recommended to use a autoloader in python, since "explicit is better than implicit" and "Readability counts".
That way, I think, I should compel the user of my module to manually import each of his/her modules, and pass these to the game instance, like:
import framework.Game
import scenarios
#many other imports
game = Game()
game.scenarios = scenarios
#so many other game.whatever = whatever
But this doesn't looks good to me, not so confortable. See, I am used to work with php, and I love the way it works with it's autoloader.
So, the first exemple has some problability to crash or be some trouble, or is it just not 'pythonic'?
note: this is NOT an web application
I wouldn't consider letting a library import things from my current path or module good style. Instead I would only expect a library to import from two places:
Absolute imports from the global modules space, like things you have installed using pip. If a library does this, this library must also be found in its install_requires=[] list
Relative imports from inside itself. Nowadays these are explicitly imported from .:
from . import bla
from .bla import blubb
This means that passing an object or module local to my current scope must always happen explicitly:
from . import scenarios
import framework
scenarios.sprites # attribute exists
game = framework.Game(scenarios=scenarios)
This allows you to do things like mock the scenarios module:
import types
import framework
# a SimpleNamespace looks like a module, as they both have attributes
scenarios = types.SimpleNamespace(sprites='a', textures='b')
scenarios.sprites # attribute exists
game = framework.Game(scenarios=scenarios)
Also you can implement a framework.utils.Scenario() class that implements a certain interface to provide sprites, maps etc. The reason being: Sprites and Maps are usually saved in separate files: What you absolutely do not want to do is look at the scenarios's __file__ attribute and start guessing around in its files. Instead implement a method that provides a unified interface to that.
class Scenario():
def __init__(self):
...
def sprites(self):
# optionally load files from some default location
# If no such things as a default location exists, throw a NotImplemented error
...
And your user-specific scenarios will derive from it and optionally overload the loading methods
import framework.utils
class Scenario(framework.utils.Scenario):
def __init__(self):
...
def sprites(self):
# this method *must* load files from location
# accessing __file__ is OK here
...
What you can also do is have framework ship its own framework.contrib.scenarios module that is used in case no scenarios= keyword arg was used (i.e. for a square default map and some colorful default textures)
from . import contrib
class Game()
def __init__(self, ..., scenarios=None, ...):
if scenarios is None:
scenarios = contrib.scenarios
self.scenarios = scenarios
A proper Python module will list all its public symbols in a list called __all__. Managing that list can be tedious, since you'll have to list each symbol twice. Surely there are better ways, probably using decorators so one would merely annotate the exported symbols as #export.
How would you write such a decorator? I'm certain there are different ways, so I'd like to see several answers with enough information that users can compare the approaches against one another.
In Is it a good practice to add names to __all__ using a decorator?, Ed L suggests the following, to be included in some utility library:
import sys
def export(fn):
"""Use a decorator to avoid retyping function/class names.
* Based on an idea by Duncan Booth:
http://groups.google.com/group/comp.lang.python/msg/11cbb03e09611b8a
* Improved via a suggestion by Dave Angel:
http://groups.google.com/group/comp.lang.python/msg/3d400fb22d8a42e1
"""
mod = sys.modules[fn.__module__]
if hasattr(mod, '__all__'):
name = fn.__name__
all_ = mod.__all__
if name not in all_:
all_.append(name)
else:
mod.__all__ = [fn.__name__]
return fn
We've adapted the name to match the other examples. With this in a local utility library, you'd simply write
from .utility import export
and then start using #export. Just one line of idiomatic Python, you can't get much simpler than this. On the downside, the module does require access to the module by using the __module__ property and the sys.modules cache, both of which may be problematic in some of the more esoteric setups (like custom import machinery, or wrapping functions from another module to create functions in this module).
The python part of the atpublic package by Barry Warsaw does something similar to this. It offers some keyword-based syntax, too, but the decorator variant relies on the same patterns used above.
This great answer by Aaron Hall suggests something very similar, with two more lines of code as it doesn't use __dict__.setdefault. It might be preferable if manipulating the module __dict__ is problematic for some reason.
You could simply declare the decorator at the module level like this:
__all__ = []
def export(obj):
__all__.append(obj.__name__)
return obj
This is perfect if you only use this in a single module. At 4 lines of code (plus probably some empty lines for typical formatting practices) it's not overly expensive to repeat this in different modules, but it does feel like code duplication in those cases.
You could define the following in some utility library:
def exporter():
all = []
def decorator(obj):
all.append(obj.__name__)
return obj
return decorator, all
export, __all__ = exporter()
export(exporter)
# possibly some other utilities, decorated with #export as well
Then inside your public library you'd do something like this:
from . import utility
export, __all__ = utility.exporter()
# start using #export
Using the library takes two lines of code here. It combines the definition of __all__ and the decorator. So people searching for one of them will find the other, thus helping readers to quickly understand your code. The above will also work in exotic environments, where the module may not be available from the sys.modules cache or where the __module__ property has been tampered with or some such.
https://github.com/russianidiot/public.py has yet another implementation of such a decorator. Its core file is currently 160 lines long! The crucial points appear to be the fact that it uses the inspect module to obtain the appropriate module based on the current call stack.
This is not a decorator approach, but provides the level of efficiency I think you're after.
https://pypi.org/project/auto-all/
You can use the two functions provided with the package to "start" and "end" capturing the module objects that you want included in the __all__ variable.
from auto_all import start_all, end_all
# Imports outside the start and end functions won't be externally availab;e.
from pathlib import Path
def a_private_function():
print("This is a private function.")
# Start defining externally accessible objects
start_all(globals())
def a_public_function():
print("This is a public function.")
# Stop defining externally accessible objects
end_all(globals())
The functions in the package are trivial (a few lines), so could be copied into your code if you want to avoid external dependencies.
While other variants are technically correct to a certain extent, one might also be sure that:
if the target module already has __all__ declared, it is handled correctly;
target appears in __all__ only once:
# utils.py
import sys
from typing import Any
def export(target: Any) -> Any:
"""
Mark a module-level object as exported.
Simplifies tracking of objects available via wildcard imports.
"""
mod = sys.modules[target.__module__]
__all__ = getattr(mod, '__all__', None)
if __all__ is None:
__all__ = []
setattr(mod, '__all__', __all__)
elif not isinstance(__all__, list):
__all__ = list(__all__)
setattr(mod, '__all__', __all__)
target_name = target.__name__
if target_name not in __all__:
__all__.append(target_name)
return target