I'm trying to keep a data science project well-organized so I've created a directory inside my src
directory called utils
that contains a file called helpers.py
, which contains some helper functions that will be used in many scripts. What is the best practice for how I should import func_name
from src/utils/helpers.py
into a file in a totally different directory, such as src/processing/clean_data.py
?
I see answers to this question, and I've implemented a solution that works, but this feels ugly:
sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.realpath(__file__))))))
Am I doing this right? Do I need to add this to every script that wants to import func_name
, like train_model.py
?
My current project folder structure:
myproject /notebooks notebook.ipynb /src /processing clean_data.py /utils helpers.py /models train_model.py __init__.py
Example files:
# clean_data.py import os import sys sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.realpath(__file__)))))) from src.utils.helpers import func_name func_name() # helpers.py def func_name(): print('I'm a helper function.')
7 Answers
Answers 1
The correct way to do it is to use __init__.py
, setup.py
and the setuptools
Python package:
myPackage/ myPackage/ __init__.py setup.py
This link has all the steps.
Answers 2
First of all, let me describe you the differences between a Python module & a Python package so that both of us are on the same page. ✌
A module is a single .py file (or files) that are imported under one import and used. ✔
import aModuleName # Here 'aModuleName' is just a regular .py file.
Whereas, a package is a collection of modules in directories that give a package hierarchy. A package contains a distinct __init__.py
file. ✔
from aPackageName import aModuleName # Here 'aPackageName` is a folder with a `__init__.py` file # and 'aModuleName', which is just a regular .py file.
Therefore, when we have a project directory named proj-dir of the following structure ⤵
proj-dir --|--__init__.py --package1 --|--__init__.py --|--module1.py --package2 --|--__init__.py --|--module2.py
๐ Notice that I've also added an empty
__init__.py
into the proj-dir itself which makes it a package too.
๐ Now, if you want to import any python object from module2 of package2 into module1 of package1, then the import statement in the file module1.py would be
from package2.module2 import object2 # if you were to import the entire module2 then, from package2 import module2
I hope this simple explanation clarifies your doubts on Python imports' mechanism and solves the problem. If not then do comment here. ๐
Answers 3
First of all let me clarify you that importing an entire module, if you are going to use a part of it, then is not a good idea. Instead of that you can use from
to import specific function under a library/package. By doing this, you make your program efficient in terms of memory and performance.
To know more refer these:
Net let us look into the solution.
Before starting off with the solution, let me clarify you the use of __init__.py
file. It just tells the python interpreter that the *.py files present there are importable which means they are modules and are/maybe a part of a package.
So, If you have N no of sub directories you have to put __init__.py
file in all those sub directories such that they can also be imported. Inside __init__.py
file you can also add some additional information like which path should be included, default functions,variables,scope,..etc. To know about these just google about __init__.py
file or take some python library and go through the same __init__.py
file to know about it. (Here lies the solution)
More Info:
So as stated by @Sushant Chaudhary your project structure should be like
proj-dir --|--__init__.py --package1 --|--__init__.py --|--module1.py --package2 --|--__init__.py --|--module2.py
So now, If I put
__init__.py
file under my directory like above, Will it be importable and work fine?yes and no.
Yes :
If you are importing the modules within that project/package directory. for example in your case
you are importing package1.module1 in pakage2.module2 as from package1 import module1
.
Here you have to import the base dir inside the sub modules, Why? the project will run fine if you are running the module from the same place. i.e: inside package2 as python module2.py, But will throw ModuleNotFoundError
If you run the module from some other directory. i.e: any other path except under package2 for example under proj-dir as python package1/module1.py
. This is what happening in your case. You are running the module from project-dir.
So How to fix this?
You have to append basedir path to system path in module1.py as
from sys import path dir_path = "/absolute/path/to/proj-dir" sys.path.insert(0, dir_path)
You have to add all the sub module paths in
__init__.py
file under proj-dir.
For example:
#__init__.py under lxml # this is a package def get_include(): """ Returns a list of header include paths (for lxml itself, libxml2 and libxslt) needed to compile C code against lxml if it was built with statically linked libraries. """ import os lxml_path = __path__[0] include_path = os.path.join(lxml_path, 'includes') includes = [include_path, lxml_path] for name in os.listdir(include_path): path = os.path.join(include_path, name) if os.path.isdir(path): includes.append(path) return includes
This is the __init__.py
file of lxml (a python library for parsing html,xml data). You can refer any __init__.py
file under any python libraries having sub modules.ex (os,sys). Here I've mentioned lxml because I thought it will be easy for you to understand. You can even check __init__.py
file under other libraries/packages. Each will have it's own way of defining the path for submodules.
No
If you are trying to import modules outside the directory. Then you have to export the module path such that other modules can find them into environment variables. This can be done directly by appending absolute path of the base dir to PYTHONPATH or to PATH.
To know more:
So to solve your problem, include the paths to all the sub modules in __init__.py
file under proj-dir and add the /absolute/path/to/proj-dir either to PYTHONPATH
or PATH
.
Hope the answer explains you about usage of __init__.py
and solves your problem.
Answers 4
On Linux, you can just add the path to the parent folder of your src
directory to ~/.local/lib/python3.6/site-packages/my_modules.pth
. See Using .pth files. You can then import modules in src
from anywhere on your system.
NB1: Replace python3.6
by any version of Python you want to use.
NB2: If you use Python2.7 (don't know for other versions), you will need to create __init__.py
(empty) files in src/
and src/utils
.
NB3: Any name.pth
file is ok for my_modules.pth
.
Answers 5
Yes, you can only import code from installed packages or from files in you working directory or subdirectories.
Answers 6
the way I see it, your problem would be solved if you would have your module or package installed, like an yother package one installs and then imports (numpy, xml, json etc.)
I also have a package I constantly use in all my projects, ulitilies, and I know it's a pain with the importing.
here is a description on how to How to package a python application to make it pip-installable: https://marthall.github.io/blog/how-to-package-a-python-app/
Answers 7
Navigate to your python installation folder
Navigate to lib
Navigate to site-packages
Make a new file called
any_thing_you_want.pth
Type
.../src/utils/helpers.py
inside that file with your favorite text editor
Note: the ellipsis before scr/utils/helpers.py
will look something like: C:/Users/blahblahblah/python_folders/scr...
<- YOU DO NEED THIS!
This is a cheap way out but it keeps code clean, and is the least complicated. The downside is, for every folder your modules are in, example.pth will need them. Upside: works with Windows all the way up to Windows 10
0 comments:
Post a Comment