[BUG] punkt_tab breaking change #3293
Description
One of the recent updates has broken the ability to obtain punkt with the traditional method:
os.system("python3 -m nltk.downloader punkt")
nltk.download('punkt')
Instead, apps which have only bumped the patch from 3.8.1 to 3.8.2 are now breaking with:
90 Resource �[93mpunkt_tab�[0m not found.91 Please use the NLTK Downloader to obtain the resource:9293 �[31m>>> import nltk94 >>> nltk.download('punkt_tab')95 �[0m96 For more information see: https://www.nltk.org/data.html9798 Attempted to load �[93mtokenizers/punkt_tab/english/�[0m
The issue is seen downstream with:
Unstructured-IO/unstructured#3511
And resolved by pinning the version to 3.8.1.