如何从另一个脚本中将命令行 arguments 传递给 python 脚本？

Question

I've got a python script that is named code.py and this takes two command line arguments --par_value xx --out_dir /path/to/dir where par_value is a certain parameter and out_dir is the output directory. I want to run this script for different values of the input parameter.

So I want to use the array:

par_vector = np.array([1,2,3])

and have the output file for each named, say, "output_1", "output_2" and "output_3". So I'm trying to write a python script where I can create that array above and an array of output strings, say, called output_strings so I can automatically run something along the like:

for i,j in zip(par_vector, output_strings): 
 python code.py --par_value i --out_dir j

But this does not run so I'm unable to figure out how to make such an automation of my code work rather than repeatedly calling the script code.py from the terminal. Any advice is welcome!

Answer 1

It might be a bit convoluted, but a way you could do it is to generate either .bat or .sh files (depending on your operating system) and call them with Popen from the subprocess library.

from subprocess import Popen, PIPE

for i,j in zip(par_vector, output_strings): 
    cmd = """python code.py --par_value {} --out_dir {}""".format(i,j)
    temp_filename = "{}{}.bat".format(i,) #swap .bat for .sh if not on Windows
    
    with open(temp_filename, 'wb') as out_file:
        out_file.write(cmd)
    
    execution = Popen([temp_filename], stdout=PIPE, stderr=PIPE)
    results = execution.communicate()
    
    # add conditions based on the output generated by results if you want to use it to verify, etc.

    os.remove(temp_filename)

Obviously this needs to be changed to match your exact needs for the file name and location, etc..

Answer 2

I would use os.system for this if you're OK with the fact that os.system() is blocking. This would look something like:

import os    
for i,j in zip(par_vector, output_strings): 
    os.system(f"python code.py --par_value {i} --out_dir {j}")