如何卸载(重新加载)Python 模块?
我有一个长期运行的 Python 服务器,并且希望能够在不重新启动服务器的情况下升级服务。 这样做的最好方法是什么?
if foo.py has changed:
unimport foo <-- How do I do this?
import foo
myfoo = foo.Foo()
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(22)
您可以使用
importlib 重新加载已导入的模块.reload()
:在 Python 2 中,
reload
是内置的。 在 Python 3 中,它被移动到imp
模块。 在 3.4 中,已弃用,取而代之的是imp
importlib
。 当目标为 3 或更高版本时,请在调用reload
时引用适当的模块或导入它。我认为这就是你想要的。 像 Django 开发服务器这样的 Web 服务器使用它,这样您就可以看到代码更改的效果,而无需重新启动服务器进程本身。
引用文档:
正如您在问题中指出的那样,如果
Foo
类驻留在foo
模块中,则必须重建Foo
对象。You can reload a module when it has already been imported by using
importlib.reload()
:In Python 2,
reload
was a builtin. In Python 3, it was moved to theimp
module. In 3.4,imp
was deprecated in favor ofimportlib
. When targeting 3 or later, either reference the appropriate module when callingreload
or import it.I think that this is what you want. Web servers like Django's development server use this so that you can see the effects of your code changes without restarting the server process itself.
To quote from the docs:
As you noted in your question, you'll have to reconstruct
Foo
objects if theFoo
class resides in thefoo
module.在 Python 3.0–3.3 中,您可以使用:
imp .reload(module)
BDFL已回答这个问题。
然而,
imp
在 3.4 中已被弃用,取而代之的是importlib
(感谢 @Stefan!)。因此,我认为您现在应该使用
importlib.reload(module)
,虽然我不确定。In Python 3.0–3.3 you would use:
imp.reload(module)
The BDFL has answered this question.
However,
imp
was deprecated in 3.4, in favour ofimportlib
(thanks @Stefan!).I think, therefore, you’d now use
importlib.reload(module)
, although I’m not sure.如果模块不是纯 Python,则删除它可能会特别困难。
以下是一些信息: 如何真正删除导入的模块?
It can be especially difficult to delete a module if it is not pure Python.
Here is some information from: How do I really delete an imported module?
reload(module)
,但前提是它是完全独立的。 如果其他任何东西都引用了该模块(或属于该模块的任何对象),那么您将收到由于旧代码停留时间比您预期的时间长而导致的微妙而奇怪的错误,以及诸如isinstance
不能跨同一代码的不同版本工作。如果您具有单向依赖关系,则还必须重新加载依赖于重新加载的模块的所有模块,以消除对旧代码的所有引用。 然后递归地重新加载依赖于重新加载的模块的模块。
如果存在循环依赖(这很常见,例如在重新加载包时),则必须一次性卸载组中的所有模块。 您无法使用
reload()
执行此操作,因为它会在刷新其依赖项之前重新导入每个模块,从而允许旧引用渗透到新模块中。在这种情况下,唯一的方法是破解 sys.modules,这是不受支持的。 您必须检查并删除您想要在下次导入时重新加载的每个
sys.modules
条目,并删除值为None
的条目来处理实现与缓存失败的相对导入有关的问题。 它不是很好,但只要您有一组完全独立的依赖项,并且不会在其代码库之外留下引用,它就是可行的。最好重新启动服务器。 :-)
reload(module)
, but only if it's completely stand-alone. If anything else has a reference to the module (or any object belonging to the module), then you'll get subtle and curious errors caused by the old code hanging around longer than you expected, and things likeisinstance
not working across different versions of the same code.If you have one-way dependencies, you must also reload all modules that depend on the reloaded module to get rid of all the references to the old code. And then reload modules that depend on the reloaded modules, recursively.
If you have circular dependencies, which is very common for example when you are dealing with reloading a package, you must unload all the modules in the group in one go. You can't do this with
reload()
because it will re-import each module before its dependencies have been refreshed, allowing old references to creep into new modules.The only way to do it in this case is to hack
sys.modules
, which is kind of unsupported. You'd have to go through and delete eachsys.modules
entry you wanted to be reloaded on next import, and also delete entries whose values areNone
to deal with an implementation issue to do with caching failed relative imports. It's not terribly nice but as long as you have a fully self-contained set of dependencies that doesn't leave references outside its codebase, it's workable.It's probably best to restart the server. :-)
对于 Python 2 使用内置函数
reload
:对于 Python 2 和 Python 3.2—3.3 使用
从模块 imp 重新加载
:对于 Python ≥3.4,
imp
已弃用支持importlib
,所以使用这个:或:
TL;DR:
Python ≥ 3.4:
importlib.reload(module)
Python 3.2 — 3.3:
imp.reload(module)
Python 2:
重新加载(模块)
For Python 2 use built-in function
reload
:For Python 2 and Python 3.2—3.3 use
reload
from module imp:For Python ≥3.4,
imp
is deprecated in favor ofimportlib
, so use this:or:
TL;DR:
Python ≥ 3.4:
importlib.reload(module)
Python 3.2 — 3.3:
imp.reload(module)
Python 2:
reload(module)
如果您不在服务器中,但在开发并且需要经常重新加载模块,这里有一个很好的提示。
首先,确保您使用的是来自 Jupyter Notebook 项目的优秀 IPython shell。 安装 Jupyter 后,您可以使用 ipython 或 jupyter console 启动它,甚至更好的是 jupyter qtconsole ,这将为您提供漂亮的彩色效果在任何操作系统中具有代码完成功能的控制台。
现在,在 shell 中输入:
现在,每次运行脚本时,您的模块都会重新加载。
除了
2
之外,还有其他选项autoreload magic:当然,它也适用于 Jupyter Notebook。
If you are not in a server, but developing and need to frequently reload a module, here's a nice tip.
First, make sure you are using the excellent IPython shell, from the Jupyter Notebook project. After installing Jupyter, you can start it with
ipython
, orjupyter console
, or even better,jupyter qtconsole
, which will give you a nice colorized console with code completion in any OS.Now in your shell, type:
Now, every time you run your script, your modules will be reloaded.
Beyond the
2
, there are other options of the autoreload magic:Sure, it will also work on a Jupyter Notebook.
以下代码允许您实现 Python 2/3 兼容性:
您可以在两个版本中将其用作
reload()
,这使事情变得更简单。The following code allows you Python 2/3 compatibility:
The you can use it as
reload()
in both versions which makes things simpler.接受的答案不处理 from X import Y 的情况。 此代码也处理它和标准导入情况:
在重新加载情况下,我们将顶级名称重新分配给存储在新重新加载的模块中的值,这会更新它们。
The accepted answer doesn't handle the from X import Y case. This code handles it and the standard import case as well:
In the reloading case, we reassign the top level names to the values stored in the newly reloaded module, which updates them.
这是重新加载模块的现代方法:
如果您想支持早于 3.5 的 Python 版本,请使用:
这定义了一个
reload
方法,可以使用模块调用该方法来重新加载它。 例如,reload(math)
将重新加载math
模块。This is the modern way of reloading a module:
If you want to support versions of Python older than 3.5, use this:
This defines a
reload
method which can be called with a module to reload it. For example,reload(math)
will reload themath
module.对于像我这样想要卸载所有模块的人(在 Emacs 下的 Python 解释器中运行时):
更多信息请参见重新加载 Python 模块。
For those like me who want to unload all modules (when running in the Python interpreter under Emacs):
More information is in Reloading Python modules.
编辑(答案 V2)
之前的解决方案仅适用于获取重置信息,但它不会更改所有引用(多于
reload
但少于所需)。 为了实际设置所有引用,我必须进入垃圾收集器,并在那里重写引用。 现在它就像一个魅力!请注意,如果 GC 关闭,或者重新加载不受 GC 监控的数据,此方法将不起作用。 如果你不想搞乱GC,原来的答案可能对你来说就足够了。
新代码:
原始答案
正如 @bobince 的答案中所写,如果另一个模块中已经存在对该模块的引用(特别是如果它是使用
as
关键字导入的,例如import numpy as np),该实例不会被覆盖。
当应用需要配置模块“全新”状态的测试时,这对我来说是一个很大的问题,因此我编写了一个名为
reset_module
的函数,它使用importlib
' sreload
函数并递归地覆盖所有声明的模块的属性。 它已经使用 Python 3.6 版本进行了测试。注意:请小心使用! 在非外围模块(例如,定义外部使用的类的模块)上使用这些可能会导致 Python 中的内部问题(例如 pickling/un-pickling 问题)。
Edit (Answer V2)
The solution from before is good for just getting the reset information, but it will not change all the references (more than
reload
but less then required). To actually set all the references as well, I had to go into the garbage collector, and rewrite the references there. Now it works like a charm!Note that this will not work if the GC is turned off, or if reloading data that's not monitored by the GC. If you don't want to mess with the GC, the original answer might be enough for you.
New code:
Original Answer
As written in @bobince's answer, if there's already a reference to that module in another module (especially if it was imported with the
as
keyword likeimport numpy as np
), that instance will not be overwritten.This proved quite problematic to me when applying tests that required a "clean-slate" state of the configuration modules, so I've written a function named
reset_module
that usesimportlib
'sreload
function and recursively overwrites all the declared module's attributes. It has been tested with Python version 3.6.Note: Use with care! Using these on non-peripheral modules (modules that define externally-used classes, for example) might lead to internal problems in Python (such as pickling/un-pickling issues).
其他选择。 请注意,Python 默认
importlib.reload
只会重新导入作为参数传递的库。 它不会重新加载您的库导入的库。 如果您更改了很多文件并且需要导入一个有些复杂的包,则必须进行深度重新加载。如果您有 IPython 或 Jupyter 安装后,您可以使用一个函数来深度重新加载所有库:
如果您没有 Jupyter,请在 shell 中使用以下命令安装它:
Other option. See that Python default
importlib.reload
will just reimport the library passed as an argument. It won't reload the libraries that your lib import. If you changed a lot of files and have a somewhat complex package to import, you must do a deep reload.If you have IPython or Jupyter installed, you can use a function to deep reload all libs:
If you don't have Jupyter, install it with this command in your shell:
那些使用 python 3 并从 importlib 重新加载的人。
如果您遇到模块无法重新加载之类的问题...那是因为它需要一些时间来重新编译 pyc(最多 60 秒)。我写此提示只是为了让您知道您是否遇到过此类问题。
Those who are using python 3 and reload from importlib.
If you have problems like it seems that module doesn't reload... That is because it needs some time to recompile pyc (up to 60 sec).I writing this hint just that you know if you have experienced this kind of problem.
Enthought Traits 有一个模块可以很好地解决这个问题。 https://traits.readthedocs.org/en/ 4.3.0/_modules/traits/util/refresh.html
它将重新加载已更改的任何模块,并更新正在使用它的其他模块和实例化对象。 它在大多数情况下不能与 __very_private__ 方法一起工作,并且可能会因类继承而阻塞,但它为我节省了大量的时间,使我在编写 PyQt guis 或运行的东西时不必重新启动主机应用程序在 Maya 或 Nuke 等程序中。 它可能在 20-30% 的时间内不起作用,但仍然非常有用。
Enthought 的包不会在文件更改时重新加载文件 - 您必须明确地调用它 - 但如果您确实需要它,那么实现起来应该不那么困难
Enthought Traits has a module that works fairly well for this. https://traits.readthedocs.org/en/4.3.0/_modules/traits/util/refresh.html
It will reload any module that has been changed, and update other modules and instanced objects that are using it. It does not work most of the time with
__very_private__
methods, and can choke on class inheritance, but it saves me crazy amounts of time from having to restart the host application when writing PyQt guis, or stuff that runs inside programs such as Maya or Nuke. It doesn't work maybe 20-30 % of the time, but it's still incredibly helpful.Enthought's package doesn't reload files the moment they change - you have to call it explicitely - but that shouldn't be all that hard to implement if you really need it
2018-02-01
foo
必须提前导入成功。from importlib import reload
,reload(foo)
31.5。 importlib — import 的实现 — Python 3.6.4 文档
2018-02-01
foo
must be imported successfully in advance.from importlib import reload
,reload(foo)
31.5. importlib — The implementation of import — Python 3.6.4 documentation
如果您遇到以下错误,此答案可能会帮助您获得解决方案:
或者
如果您有如下所示的导入,则可能需要使用 sys.modules 来获取要重新加载的模块:
主要问题是 importlib.reload code> 仅接受模块而不接受字符串。
If you encounter the following error, this answer may help you to get a solution:
OR
In case you have an import like the one below, you may need to use the
sys.modules
to get the module you want to reload:The main issue is that the
importlib.reload
accepts module only not string.对我来说,Abaqus 就是这样工作的。
假设您的文件是 Class_VerticesEdges.py
for me for case of Abaqus it is the way it works.
Imagine your file is Class_VerticesEdges.py
当
重新加载
时,Python不会重新计算子模块地址,如果它位于sys.modules
中,这是一个解决方法,并不完美,但有效。
Python will not re-calculate submodule address when
reload
, event if it is insys.modules
Here is a workaround, not perfect but worked.
从 sys.modules 中删除模块还需要删除'None'类型。
方法 1:
方法 2,使用簿记条目,删除所有依赖项:
可选,只是为了确定所有条目均已退出,如果您这样选择:
Removing modules from sys.modules requires 'None' types to be deleted as well.
Method 1:
Method 2, using bookkeeping entries, to remove all dependencies:
Optional, just to be certain all entries are out, if you so choose:
我在尝试在 Sublime Text 中重新加载某些内容时遇到了很多麻烦,但最终我可以编写这个实用程序,根据
sublime_plugin.py
用于重新加载模块的代码在 Sublime Text 上重新加载模块。下面允许您从名称中带有空格的路径重新加载模块,然后在重新加载后您可以像平常一样导入。
如果您第一次运行,这应该加载模块,但如果稍后您可以再次使用方法/函数
run_tests()
,它将重新加载测试文件。 对于 Sublime Text (Python 3.3.6
),这种情况经常发生,因为它的解释器永远不会关闭(除非您重新启动 Sublime Text,即Python3.3
解释器)。I got a lot of trouble trying to reload something inside Sublime Text, but finally I could wrote this utility to reload modules on Sublime Text based on the code
sublime_plugin.py
uses to reload modules.This below accepts you to reload modules from paths with spaces on their names, then later after reloading you can just import as you usually do.
If you run for the first time, this should load the module, but if later you can again the method/function
run_tests()
it will reload the tests files. With Sublime Text (Python 3.3.6
) this happens a lot because its interpreter never closes (unless you restart Sublime Text, i.e., thePython3.3
interpreter).另一种方法可能是在函数中导入模块。 这样,当函数完成时,模块就会被垃圾收集。
Another way could be to import the module in a function. This way when the function completes the module gets garbage collected.