Dependency on Numpy not declared in setup.py

[nluetts]

Hi,

thanks for developing HAPI 😄

I just tried to install HAPI into a new virtual environment and it failed with the message:

ModuleNotFoundError: No module named 'numpy'

This is because Numpy is not declared in setup.py as a dependency.
You can use the install_requires keyword argument to include Numpy as an explicit dependency:

setup(
    name='hitran-api',
    version=HAPI_VERSION,
    packages=['hapi',],
    install_requires=["numpy",], # This is missing. Better specify a version, e.g. "numpy>=1.20" or whatever HAPI needs
    license='MIT',
)

The next problem is that setup.py imports HAPI_VERSION and HAPI_HISTORY form hapi.hapi.
This leads to setup.py trying to import Numpy, which then fails, although Numpy is now a declared dependency.
I solved this problem by moving the declarations of HAPI_VERSION and HAPI_HISTORY into a file _version.py in the root of the repository and then did

from _version import HAPI_VERSION, HAPI_HISTORY

in setup.py.

If you want to keep these declarations inside the hapi folder, I guess you would need to remove

from .hapi import *

from __init__.py, because this triggers the import of Numpy.

You should be able to do this if you instead declare the exported functions with the __all__ variable, like described here:
https://docs.python.org/3/tutorial/modules.html#importing-from-a-package

This problem seemed to be undetected for quite a while, I guess because most people have Numpy installed.
However, if your want to use a dedicated environment for your project (like I always do), Numpy will not be available by default.

Best regards
Nils

[nluetts]

Oh, moving HAPI_VERSION and HAPI_HISTORY into the root folder does not work, because it is needed in the hapi module itself. The dirty and not well maintainable hack would be to just have a copy of _version.py in the hapi folder, but I guess using __all__ is the better solution. You seem to have used __all__ previously, it is commented out in __init__.py. Was there a problem?