I'm writing a program for my grandpa. Occasionally an update will require a new module that's not installed on his computer. I don't want to have to manually install new modules every time, but he's not tech savvy enough to figure it out himself.
Is it OK for me to check for ImportErrors and install a module from the error message? If not, what's the alternative?
This is the job of the package manager. In the case of Python: pip.
If you package the library you are building and specify its dependencies, your users (grandpa), will simply need to run:
$ pip install --upgrade <name-of-package>
to get the latest version you've released and all of its dependencies. If you don't want to publish your code online, pip can also install from local sources:
$ pip install zipped_package_emailed_to_grandpa.zip
or, if you really need to package this in a single (Python 3.4+) script, you could call pip from within:
#!/usr/bin/env python3
def main():
...
if __name__ == '__main__':
import sys, subprocess
dependencies = ['dep_1', 'dep_2']
subprocess.call([sys.executable, '-m', 'pip', 'install'] + dependencies)
import dep_1, dep_2
main()
but this is definitely uncommon.
pip is something you run "simply".
pip can read the package list from a file - pip install -r requirements.txt. The OP can distribute requirements.txt as part of their program, and when they add a new module requirement they add it to requirements.txt. The installation command can be wrapped in a shell script that the grandpa just need to double-click.
Commented
Jul 1, 2015 at 19:56