bug/_download_nltk_packages_if_not_present throws HTTP 403 Forbidden

[tn-halfspace]

Describe the bug
Until 25.11.2024 I haven't seen any problems with this function. Since now this url:

https://utic-public-cf.s3.amazonaws.com/nltk_data_3.8.2.tar.gz

Returned 403.

Edit: it's back up there... But seems inconsistent.

[villagab4]

Also experiencing this issue !

[derekhsu]

I still have the issue now. Where can I download and install it?

[akhilkapil]

Hi @tn-halfspace, were you able to resolve this issue?

[tn-halfspace]

This is ridiculous @Unstructured-DevOps. It hapenned again today.

I've created my own workaround to use nltk's downloader directly instead of relying on unstructured.nlp downloader and download the two packages that are used by unstructured.io (averaged_perceptron_tagger_eng and punkt_tab)

I think in a perfect world the best practice would be to ship your code with these two packages and use manual installation by changing the NLTK_PATH according to NLTK documentation

[vangheem]

If you upgrade to the latest, it should be fixed: #3796

[skshetry]

I am still getting AccessDenied. @vangheem, would it be possible to keep the URL up?

We are still stuck with an older version of unstructured due to #3731 at the moment.

[kartik4949]

Following could be a hack for solving this issue.

import nltk
nltk.download('punkt')
nltk.download('averaged_perceptron_tagger')

[skshetry]

Following could be a hack for solving this issue.

import nltk
nltk.download('punkt')
nltk.download('averaged_perceptron_tagger')

yeah, that's what I am doing right now (iterative/datachain#687). :)

[djibomar]

This started happening for us today as well. The previous version used to call this s3 url to download nltk https://utic-public-cf.s3.amazonaws.com/nltk_data_3.8.2.tar.gz however it now just throws a 403.
Why did this happen ?

[jexp]

Same here, we're also seeing the 403 errors.

[zya]

I'm experiencing this intermittently as well.

[dipanjanS]

This is still happening and has broken all our pipelines

[jan-schneider3]

Same here

[zya]

Updating to 0.16.11 seems to have addressed the issue for me.

[dipanjanS]

Still doesnt work even with 0.16.11, literally the same issue but 0.15 seems to be working, this is really frustrating

[dipanjanS]

Reinstalled everything and reupdated, its working now with the latest version so far, however I have other issues now where the outputs have changed :) but that's for another day (issue). Thanks for all the suggestions in this thread!

[doyeka]

If you upgrade to the latest, it should be fixed: #3796

Can confirm upgrading to the latest version of unstructured (0.16.11) resolved the bug for me.

[scanny]

Closing as resolved. If you're still having trouble feel free to reopen.