Question

Currently working with the following package structure:

/package
    __init__.py
    final.py
    /write
        __init__.py
        write.py
    /data
        backup.txt
        backup1.txt
        backup2.txt

final.py imports write.py, which should be able to go back one directory and write a series of backup .txt files to /data.

final.py should be able to go into /data during another call and access the backup files, hence the need to save the information in /data.

1. I’m not sure this should be the correct hierarchy for a package file? How would /write create text files in a directory branch separate from itself without using absolute file paths in case the whole project file is moved, say onto a server.

2. Would it be wrong (once the backup.txt files are created) to add a retrieve.py to /data which returns the .txt files in some sort of data structure and make /data a package, or (2) should final.py directly enter /data and retrieve the text files.

Answer 1

1. IMO you shouldn’t be writing into your packages. Set your code up so that you are writing to a data directory that is potentially outside your package. Numerous code deployment strategies assume that your code will be in a directory that is not normally writable. (E.g. if it is packaged for common linux distributions, the code will go into /usr/lib/python…/yourpackage/ and the data will be written to /var/lib/yourpackage, or something similar.)
2. Put your retrieve.py outside of …/data, possibly in a …/read directory, or alongside final.py, depending on the organization you need.

To write to an arbitrary location, just pass the full path to open. For example, assume that you store the path to your data directory in a constant:

DATA_PATH = '/var/lib/mypackage'

def backup():
    f = open(os.path.join(DATA_PATH, 'backup.txt'), 'w')
    f.write('some backup data...')
    f.close()