如何在不重复导入顶级名称的情况下构造python包
问题描述
我是 python 包管理的新手，肯定做错了什么.我被鼓励创建一个目录结构如下:
I'm brand new at python package management, and surely have done something wrong. I was encouraged to create a directory structure as follows:
bagoftricks
├── bagoftricks
│ ├── bagoftricks
│ │ ├── __init__.py
│ │ └── bagoftricks.py
│ └── __init__.py
├── README.md
└── setup.py
bagoftricks.py 包含两个函数，levenshtein() 和 geofind().
bagoftricks.py contains two functions, levenshtein() and geofind().
我想把它们称为:
import bagoftricks
x = bagoftricks.levenshtein(arg1,arg2)
相反，我发现我必须这样做:
Instead, I find I have to do this:
import bagoftricks
x = bagoftricks.bagoftricks.levenshtein(arg1,arg2)
有没有更好的方法来组织我的包裹，而不用重复命名?
Is there a better way to organize my packages in the first place, without the naming redundancy?
更新
所以，我按照下面 Avichal Badaya 的说明，移除了一层嵌套.也就是说，我现在有……
So, I followed Avichal Badaya's instructions below, and removed one level of nesting. That is, I now have...
bagoftricks
├── bagoftricks
│ ├── __init__.py
│ └── bagoftricks.py
├── README.md
└── setup.py
但是，要调用这个包，我还是有...
However, to call this package, I still have...
from bagoftricks.bagoftricks import geofind()
或
import bagoftricks
然后
>>> bagoftricks.bagoftricks.geofind()
而不是想要的......
Rather than the desired....
from bagoftricks import geofind()
或
import bagoftricks
>>> bagoftricks.geofind()
我无法移除额外的嵌套层.以此类推，当我尝试移除一层嵌套时，我的模块是扁平的，如下所示:
I cannot remove that extra layer of nesting. When I try, by analogy, to remove one more level of nesting, so that my module is flat, as:
bagoftricks
├── __init__.py
├── bagoftricks.py
├── README.md
└── setup.py
我根本无法构建包...
I cannot build the package at all...
$ python setup.py build
running build
running build_py
error: package directory 'bagoftricks' does not exist
像标准包一样使用自然导入，没有多余的顶级名称导入的秘诀是什么?
What's the secret for natural imports like standard packages use, without redundant top-level name imports?
解决方案
第一级bagoftricks"就可以了.可以这么说，这只是您的项目"的名称.在你有一个 setup.py 和其他文件告诉打包系统他们需要知道什么.
The first level "bagoftricks" is fine. That's just the name of your "project" so to speak. In the you have a setup.py, and other files that tell the packaging systems what they need to know.
然后您可以将代码直接放在该模块中，或者放在 src 目录中.您甚至可以只使用这种结构:
You can then have the code directly in this module, or in a src directory. You can even go as far as just having this structure:
bagoftricks
├── bagoftricks.py
├── README.md
└── setup.py
但我不建议这样做，主要是因为您可能想稍后重新组织，如果您已经有一个合适的"包会更容易.此外，大多数人、工具和文档都假设你有一个包，所以它更容易.
But I would not recommend that, mostly because you might want to reorganize things later, and it's easier if you already have a "proper" package. Also most people, tools and docs assume you have a package, so it's easier.
所以最小值是:
bagoftricks
├── bagoftricks
│ └── __init__.py
├── README.md
└── setup.py
使用 __init__.py 包含您要导入的函数.然后你可以像这样使用这些函数:
With __init__.py containing the functions you want to import. You then use these functions like this:
from bagoftricks import levenshtein, anotherfunction
一旦 __init__.py 变得太大，你想把它分成几个模块，给你这样的东西:
Once that __init__.py becomes too big, you want to split it up in several modules, giving you something like this:
bagoftricks
├── bagoftricks
│ ├── __init__.py
│ ├── anothermodule.py
│ └── levenshtein.py
├── README.md
└── setup.py
您的 __init__.py 然后应该从各个模块导入函数:
Your __init__.py should then import the functions from the various modules:
from bagoftricks.levenshtein import levenshtein
from bagoftricks.anothermodule import anotherfunction
然后你仍然可以像以前一样使用它们.
And then you can still use them like like you did before.
相关文章