DOC: Fix up auto build procedure to upload Cheat Sheet to web site

[Dr-Irv]

Need to fix the procedures so that the PDFs (English, Japanese) for the cheat sheet here: https://github.com/pandas-dev/pandas/tree/master/doc/cheatsheet

get uploaded to here:
https://pandas.pydata.org/Pandas_Cheat_Sheet.pdf

(and when the Japanese one is there, can update the docs to link to it as well)

Note: If we decide to move the location on the web site, then we should do a redirect, because there are links out there to the cheat sheet in the current location.

Followup to discussion here: #40168 (comment)

[jorisvandenbossche]

For the website and docs, we rsync to the server for every commit on master. For example for the website:

pandas/.github/workflows/ci.yml

Lines 124 to 126 in a25198c

	- name: Upload web
	run: rsync -az --delete --exclude='pandas-docs' --exclude='docs' --exclude='Pandas_Cheat_Sheet*' web/build/ docs@${{ secrets.server_ip }}:/usr/share/nginx/pandas
	if: github.event_name == 'push'

I assume the reason the --exclude='Pandas_Cheat_Sheet*' is there, is because the cheatsheet is not actually included in the built web sources, so when rsyncing that, the cheatsheet would otherwise be removed.

But adding an additional rsync call to synchronize the cheatsheets should be possible?

(cc @TomAugspurger @datapythonista)

[datapythonista]

Instead of another rsync, I think we can simply cp the cheat sheets from the repo to right location in the built web / docs after the web / docs are built. Does it make sense?

[jorisvandenbossche]

Yes, that's indeed even easier.

[Hallicopter]

I can give this a shot, seems like a good starter issue for a new contributor like me :)
From what I understand of the issue, I would have to make a change in the CI file pointed out here, so that as a part of the build pipeline the cheatsheet files are also cp'd in, correct?

I will have to figure out the Github workflows part, but I think I can try for a PR by the weekend. Is there anything else I should know? (Especially as a new contributor to this project?)

[Hallicopter]

take

[Hallicopter]

I think we can simply cp the cheat sheets from the repo to right location in the built web / docs after the web / docs are built. Does it make sense?

I went through the contribution guidelines and got a good sense of the overall build workflow. I am still a little confused as to where should I cp the cheatsheets to? Ie., where is the right location in the built web/docs that these cheatsheets can be copied to so that it shows up https://pandas.pydata.org/Pandas_Cheat_Sheet.pdf

Sorry if I am asking obvious questions, but I am unfamiliar with the website build workflows for Pandas overall, so asking what comes to my mind.

Would adding the following solve the problem (again, unsure of the destination for the cp)

- name: Move cheatsheets into site directory
  run: cp doc/cheatsheets/Pandas_Cheat_Sheet* web/build/docs

(this would be added right above name: Save website as an artifact)

Correct me if I have misunderstood anywhere.

[jorisvandenbossche]

I am still a little confused as to where should I cp the cheatsheets to? Ie., where is the right location in the built web/docs that these cheatsheets can be copied to so that it shows up https://pandas.pydata.org/Pandas_Cheat_Sheet.pdf

Since the cheatsheet uses a "top-level" URL (it lives directly under pandas.pydata.org), I think it should be copied to web/build/, as that is the main directory that is synced with the server for the main website.

Would adding the following solve the problem

Yes, that seems correct to me (with web/build as target then)

@Hallicopter Thanks for working on it!

[datapythonista]

Not sure, but just came to my mind that maybe we have a nginx rule for the cheatsheets. I remember we had something being served from a different place, and would make sense that are the cheatsheets, otherwise they could be overwritten when updating the web/docs.

I'm in a bus the whole day, so can't check until tomorrow. But if that's the case we'll have to synchronize merging the PR with updating nginx.

[Hallicopter]

Yep, no problem! If I were to locally test my changes, is there any suggestion you have? I can just build the website locally using the commands I am assuming.
Also, if there are some other nuances regarding the nginx configs, do let me know!

[datapythonista]

I was just checking, there is nothing in the nginx config. It's just the exclude in the rsync that is being used to avoid them being overwritten.

I'm +1 on moving the cheatsheets to their own directory (e.g. cheatsheets/).

In any case, your command seems right (I'd just add Copy instead of Move in the title since that's what you're doing). And then you'll have to remove the exclude of the cheatsheets in the sync.

[Hallicopter]

That's great! I shall put in a PR in a few days for review.

Also, for my understanding here

--exclude='pandas-docs' --exclude='docs' --exclude='Pandas_Cheat_Sheet*'

Why are pandas-docs and docs excluded?

[Dr-Irv]

@Hallicopter any update on doing a PR for this?

[trallard]

take