How to build a wheel for nltk which includes wordnet (no download required)? - nltk

Question

I am building a wheel for a package that uses word-forms which in turn uses nltk. When this package runs and imports word-forms that package then calls nltk.download("wordnet") which then croaks since it's running on a machine behind a firewall.
For example:
>>> from word_forms.word_forms import get_word_forms
[nltk_data] Error loading wordnet: <urlopen error [WinError 10060] A
[nltk_data] connection attempt failed because the connected party
[nltk_data] did not properly respond after a period of time, or
[nltk_data] established connection failed because connected host
[nltk_data] has failed to respond>
>>> import nltk
>>> nltk.download('popular')
[nltk_data] Error loading popular: <urlopen error [WinError 10060] A
[nltk_data] connection attempt failed because the connected party
[nltk_data] did not properly respond after a period of time, or
[nltk_data] established connection failed because connected host
[nltk_data] has failed to respond>
False
My idea is to build a wheel for nltk which already includes the wordnet dataset, assuming that this will bypass the nltk.download() call. Or maybe instead hack the word-forms code to somehow accomplish the same thing? I will build the wheel file on a machine that does have access to the internet and then copy it into the firewalled machine via SFTP.

Answer 1

I have fixed this by downloading the NLTK data on a separate machine then uploading it to the firewalled server via SFTP.
The download was performed using the command python -m nltk.downloader all.
Then in my Python code on the firewalled machine, I have added this:
os.environ['NLTK_DATA'] = 'H:/path/to/nltk_data'
Everything seems to work as advertised now.
(Thanks to the word_forms project for the helpful guidance on this issue.)