How to pass argv inputs from one python script with parameters to one without?

I am dealing with large sensor data structures stored in FITS-files which I open and read-in using the astropy module. For analysis I have written several scripts (e.g. analysis1.py) that deal with this data in different ways - each one creating an image file with several subplots.

Now I want to create an init.py script that initializes things like, filename and several parameters for analysis like this

init.py filename arg2 arg3 arg4

This I want to pass then to all subscripts, e.g. analysis1.py

#analysis1.py

from init import filename, arg2, arg3, arg4

# import everything else necessary

# do things with the data
# output picture as png-file

using bash but without explicitely inputting these as arguments again like

#!/bin/bash
# initialization of parameters and data read-in
python init.py filename arg2 arg3 arg4

# analysis scripts running in parallel using the parameters of init.py
python analysis1.py &
python analysis3.py &

I use this way because I'm usually working on CPU-machines and want to use as many threads/cores as possible to be efficient.

Common importing of course doesn't work because there argv doesn't get the parameters from the last run

import init
from sys import argv

print(argv)
#returns ['mypath/init.py']

How could I achieve this?

Solution

As Kendas says, you will have to persist the arguments on disk somewhere.

This will work:

init.py

import sys
import pickle

def get_parms():
    return pickle.load(open("init.dat","rb"))

if __name__ == "__main__":
    myparms = sys.argv[:]
    pickle.dump(myparms,open("init.dat","wb"))

analysis1.py

import init
myparms = init.get_parms()
print(myparms)

If you run init.py like this:

python init.py filename arg1 arg2 arg3

then analysis1.py will produce this output:

['(pathname)/init.py', 'filename', 'arg1', 'arg2', 'arg3']

You will almost certainly want to allow analysis1.py to accept command line parameters as overrides to the values from init.