Fix broken documentation link to mycorpus.txt

[benbroks]

Fixes #2872

Motivation: Old notebook was dependent on a URL for a text file. Wanted to make sure this was able to run locally. No has to download to run the notebook.

Fixes: Creates mycorpus.txt inline. Fixed an image reference at the end of the same notebook.

[piskvorky]

@mpenkov this file is auto-generated, right? What was the process for updating the existing docs?

[gojomo]

Given that the file is only ~19 lines and ~69 words, I'm surprised it was ever set up to be downloaded from a remote URL!

[piskvorky]

It was a demonstration of "remote input streaming" IIRC. That's what the surrounding text explains.

But not worth the trouble, probably. I put the file back now, in any case.

[mpenkov]

Yes, the .ipynb is an autogenerated file. We should not be changing it directly. If we want to go ahead with this update, we have to update the .py file in docs/src/gallery.

Personally, I'm not sure of what the benefit of ensuring that notebook runs without a network connection out of the box, but if we do ahead with it, then:

1. We should update the .py file and generate the .ipynb from it
2. We don't need to import smart_open anymore.
3. We don't need to write the corpus to a file. The generator can just yield the strings as necessary.

[benbroks]

Apologies if this isn't along the lines of what you were looking for. Made the same changes in the .py file.

[mpenkov]

@benbroks Sorry for the delay in reviewing this.

It looks like the documentation build is failing. Is it working for you locally?

If yes, then it's likely you haven't included all autogenerated files in this PR. At a glance, gensim/docs/src/auto_examples/core/run_corpora_and_vector_spaces.rst should also be included.

If no, then please make sure the docs build locally, and then update this PR.

[piskvorky]

Ping @mpenkov this issue keeps coming back, people following the tutorial still get tripped up.

I placed the file under https://radimrehurek.com/mycorpus.txt, instead of https://radimrehurek.com/gensim/mycorpus.txt, so it doesn't get overwritten / deleted with each release. Do we just update the location in the tutorial(s)?

I'm not a fan of the approach from this PR – not portable (/tmp/…) and adds too much distracting setup (read/write files, irrelevant).

[mpenkov]

We don't need to do anything with this PR, this is already resolved by #3148

@benbroks Thank you for your contribution, but the problem has already been resolved by another PR.

The public-facing documentation will get updated as soon as we do the next release.