Python project structure

Question

I've started a Python project. It's not a web app or GUI app, just a simple command line tool. I generally use the same kind of project structure, which is:

tspviz
│   .gitignore
│   requirements.txt
│   setup.py
│
├───data
│       prices.txt
│       shops.txt
│
├───tests
└───tspviz
    │   tspviz.py
    │   __init__.py
    │
    ├───algorithm
    │   │   genetic.py
    │   │   params.py
    │   │   __init__.py
    │   │
    │   └───__pycache__
    │           genetic.cpython-37.pyc
    │           params.cpython-37.pyc
    │           __init__.cpython-37.pyc

So it is quite self-explanatory, the main file is the tspviz.py in tspviz directory. Now this is a program which will solve the Travelling Salesman Problem, so I would need to make some classes like City, Path etc. Then the problem is: where do I put them? I would make a separate directory in the tspviz dir (not the root one). Then, well, how do I name this folder without confusing my colleagues? I don't know, "classes", "types", "misc"? So my question would be: Should I place them all in a directory inside tspviz as they would probably be used only from there and secondly: is there a standardized way to name a folder like this? (with classes, that you use all over the place)

bruno desthuilliers · Accepted Answer · 2018-08-29 11:24:02Z

0

given the classes names (City, Path etc) it looks like your classes belong to the problem domain, so a models.py module might make sense. Note that you don't necessarily need a folder here, Python is not Java and doesn't forces you into this "one file per class" non-sense - only use a folder (package) if you have other reasons to split your module in submodules.

As to "where to put it" (wheter a module or package), well, at the same level as your other modules and packages obviously (in the tspviz/tspviz subfolder, along tspviz.py and the algorithm package).

is there a standardized way to name a folder like this?

No.

(with classes, that you use all over the place)

Python's classes are objects, just like functions and everything else, so the fact there are (or not) classes defined in your module / package is totally irrelevant, it would be the same with functions etc.

edited Aug 29, 2018 at 11:24

answered Aug 29, 2018 at 11:18

bruno desthuilliers

78.3k6 gold badges103 silver badges129 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

minecraftplayer1234 Over a year ago

Okay, that's great, thank you. And do you think I should actually get rid of the algorithm folder, where genetic.py defines a class to solve TSP with genetic algorithm and the params.py is just a file with one dataclass?

bruno desthuilliers Over a year ago

@Frynio not necessarily. There's no "one size fits all" solution and no golden rule here, except the usual ones: good naming, high cohesion and a low coupling. oh and yes, the technical fact that imported modules are only recompiled (at import time) if the .pyc is outdated, while the main script is always recompiled, so the less code in your main script the better (wrt/ startup time I mean).

minecraftplayer1234 Over a year ago

Oh wow, I actually didn't know that. So I could potentially in main just do if __name__==..., the run some function from other imported file, and that's how I actually gain some time, right?

bruno desthuilliers Over a year ago

yes exactly - your program's main script should only import the minimal required modules, parse arguments, setup things like logging, configuration etc and run. Everything else belongs to modules/packages. Well, unless it's a small utility script you want to keep as standalone, of course.

Collectives™ on Stack Overflow

Python project structure

1 Answer 1

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related