Add a sample ZIM file suitable for automated unit & UI tests in kiwix-js - Githubissues

openzim / zim-testing-suite

This repository contains testing zim files for libzim and other openzim repositories.

3 stars 3 forks source link

Add a sample ZIM file suitable for automated unit & UI tests in kiwix-js #5

Open mossroy opened 3 years ago

mossroy commented 3 years ago

In kiwix-js, we currently use an old wikipedia_en_ray_charles_2015-06.zim to run our automated unit and UI tests. We would need an equivalent more recent one (with no-namespace, zstd, webp etc) so that these new formats/contents are covered by CI tests.

I see that the same content has been updated very recently with https://download.kiwix.org/zim/wikipedia/wikipedia_en_ray_charles_maxi_2021-08.zim . It seems to use zstd and webp, but not the no-namespace format.

Could this Ray Charles content be updated with no-namespace format? It would be ideal for us as we could keep most of our test scenario.

If it's not possible, another content might be used instead, with similar criteria :

wikimedia-based
with some images (and other types of content if possible, like audio) inside articles
with a main page
with some hyperlinks between articles
with all the recent OpenZIM features and contents
small file (ideally around 1MB)
split in several .zimxx files (but we'll do that ourself)

I did not find such file in https://github.com/openzim/zim-testing-suite/tree/main/data/nons (wikibooks and the "climate change" one do not seem to have images inside articles)

mossroy commented 3 years ago

I see there's no newer version of the Ray Charles archive, yet. Do you think we could have a no-namespace version of it, in a near future?

kelson42 commented 3 years ago

I think this will take 1-2 months.

mossroy commented 2 years ago

Is latest Ray Charles archive https://download.kiwix.org/zim/wikipedia/wikipedia_en_ray_charles_maxi_2022-01.zim suitable for our tests? It seems to me that it's still not using the no-namespace scheme?

kelson42 commented 2 years ago

@mossroy MWoffliner is still not updated to libzim7

mossroy commented 2 years ago

OK. When will it be, so that we have a no-namespace version of the Ray Charles ZIM file?

kelson42 commented 2 years ago

I can not answer to that question, work is ongoing.

mossroy commented 2 years ago

Sorry if my question seemed rude, it was not my intention. I understand things can not always be done as initially planned, and I know everybody does his best.

I was asking because you gave us a supposed timeline in a previous comment of this issue. Is there a blocker that has been discovered in the meantime?

kelson42 commented 2 years ago

@mossroy No problem, I was just too optimistic last September and now I just try to be right. Most of the work on the side of the node-libzim seems to have been done, but I can hardly tell more. I would not rely on this for anything important.

mossroy commented 2 years ago

Thanks for the info. Our goal is to have automated testing on ZIM files that are representative. We currently have automated testing on a very old Ray Charles archive, which is interesting for backward compatibility. But it does not cover all the more recent features we have in ZIM files. Is it important? I think so. Good test coverage is a key to secure PRs and releases.

Ray Charles archives seem to be the best candidates for our automated testing. I'm not sure if there are other ones that would match the criteria listed above?

mossroy commented 2 years ago

It looks like https://download.kiwix.org/zim/wikipedia/wikipedia_en_ray_charles_nopic_2022-05.zim is still not using the no-namespace scheme. Any ETA, so that we can finally implement suitable unit tests in kiwix-js?