正确的方式来组织python项目结构和分发

Question

I have a Python client-service project. Logical dependencies look like this:

Say in words, I have some services ( srv1 , srv2 , ..., srvX ) with common code in service.py , some clients for these services ( srv1_cl.py , srv2_cl.py , ..., srvX_cl.py ) with common code in client.py and common code for both service and clients in common.py .

Now I have problem with structuring my git repository for development and deploying my project to users.

Firstly I tried to store my clients and service in separate folders. Moreover, for each service I created a folder, because each service can contain a lot of files.

/myrepo.git
    common.py
    /clients
        client.py
        srv1_cl.py
        srv2_cl.py
    /services
        service.py
        /srv1
            conf.py
            srv1.py
            start.bat
        /srv2
            conf.py
            srv2.py

When I store my files like that I ran into a problem - how can a Python script export files from upper directories? Because I need test my client scripts, make some changes and execute.

I found some solutions based on adding PATH with upper folders ( sys.path.insert(0, parentdir) , but I'm sure it's not a good way) and relative imports . But for relative imports I must wrap my scripts into one package.

OK, let's say I will do this, but here's another problem. I want to deploy "clients" separately from "services" because I don't want share the services' code with my users. How can I handle this problem?

What is the best structure for my project? How can I develop and debug my project when I wrap my code to a package? Because for testing I want to start clients from my subfolders.

Answer 1

Good question. I ran into a similar problem with importing modules at different levels in the directory tree, and deployment. I ended up up reorganizing the project tree to be very flat(all files in one directory) and just naming the files informativeley, and refractoring code a lot to get the desired functionality. And the for deployment, I copied the main pyc file into the subproject folder. I would love to know a better way though.