Use flat folder for JATS output

[tarleb]

No description provided.

[tarleb]

@xuanxu I forgot, how are the artifacts copied to the target repository? Would it help if Inara took a target directory as another argument?

[xuanxu]

@tarleb all files in the media folder are copied to the target repository (also to a media folder).
It is useful to get all media files from Inara in the media folder to know what to copy.
If we don't want the files to be in a final folder, I think is better to update it in the action pushing to the repo than here.

[xuanxu]

If I understand it correctly, we need:

• All the images in the same directory as the jats file
• The correct relative references in the JATS file to the images
• A directory to know what to upload to the final repository

Would it be easier to have jats file and images in a separate jats_files (for example) directory?

[arfon]

Would it be easier to have jats file and images in a separate jats_files (for example) directory?

Yeah, I agree I think this would be the ideal solution here (a self-contained folder with everything in related to JATS, but also without any nested folders).

[tarleb]

I think this should do for now: Instead of a file paper.jats, we now create a folder paper.jats that has all the relevant data, including the images, all at the same level. Example:

paper.jats
├── mandrill.jpg
├── nyan-cat.png
├── paper.jats
└── sylt.jpg

[tarleb]

If that's easier, then we could also make that a paper.jats.zip archive.

[arfon]

we now create a folder paper.jats that has all the relevant data, including the images, all at the same level. Example:

Just for absolute clarity – the paper (e.g., paper.jats.xml) will also be in that same directory?

[xuanxu]

I've prepared a branch in the GH actions to use this changes and keep the paper.jats folder when uploading to the papers repo. I'll merge the changes once this new version of inara is up.

[xuanxu]

Just for absolute clarity – the paper (e.g., paper.jats.xml) will also be in that same directory?

Yes, the paper.jats folder includes the xml and any media files

[tarleb]

Sorry, I'm not entirely clear on these questions:

1. Is there more work needed on Inara?
2. Who will merge the PRs? (I assume @xuanxu?)

[xuanxu]

1. Is there more work needed on Inara?

I don't think so (every new commit is automatically re-publishing the Inara Docker image, right?). This is ready.

2. Who will merge the PRs? (I assume @xuanxu?)

I'll merge them all at the same time tomorrow morning so I can monitor the actions.

[arfon]

Thanks @tarleb!

[xuanxu]

Looking good!
https://github.com/openjournals/joss-papers/tree/master/joss.06017
https://github.com/openjournals/joss-papers/tree/joss.05618/joss.05618

[arfon]

Hang on, isn't this URL incorrect?

https://github.com/openjournals/joss-papers/blob/eff5cbbffd13961e4399e80bde0033daed1221de/joss.05618/paper.jats/10.21105.joss.05618.jats#L159

(i.e., I think it should be ./matbench-genmetrics.png)

[xuanxu]

@arfon ouch, you are right, the files are in the right place but the URLs for images in the paper all point to /paper.jats/ folder https://github.com/openjournals/joss-papers/blob/master/joss.06017/paper.jats/10.21105.joss.06017.jats#L253
Is this easy to fix @tarleb?

[tarleb]

Argh, my bad. I'll open a new PR in a minute.

[tarleb]

See #53. Do we want to keep the /paper.jats/ path or shorten that to just /jats/?

[xuanxu]

Merged! I think paper.jats is fine

[xuanxu]

@arfon can you reaccept this one to regenerate the files? openjournals/joss-reviews#6017

[arfon]

Done

[xuanxu]

Hey @tarleb, merging #53 seems to have modified the DTD declaration in the generated JATS file, and that makes the validation fail, see the first lines here:
openjournals/joss-papers@554242d
Is that intentional (in which case I'll update the xml validator) or is a bug?

[xuanxu]

I've updated the XML validator gem so now we can validate both the Journal Publishing and the Journal Archiving and Interchange tag sets.

Still not sure which of them we should use in Inara to generate the paper.jats file.
@arfon I think PubMed recommends the Journal Publishing one?

[tarleb]

This was not intended, sorry. I'll push the fix without PR.

Note to self: better not to rush out patches at 2am after the ESC.

[xuanxu]

Do we want to keep the /paper.jats/ path or shorten that to just /jats/?

hey @tarleb, I noticed that sometimes there's already a paper.jats file in the repository created by the authors of the submission and that has caused conflicts on a couple reviews already, so on a second thought it's better to use another name for the folder, /jats/ or /paper_jats/ for instance.

[tarleb]

I've opened #54 for that.

[arfon]

Hey both. Can I just check where we landed here? In the joss-papers repo I see a PR that's flattening folders (good), but doesn't seem to be putting the papers into a jats folder.

Also, this most recently-merged PR looks to still be using the old paper.jats repo format?

[xuanxu]

Right now, in the joss-papers repo all the jats files (.jats and images) are placed in a paper.jats folder, so the structure is something like this:

joss-paper-1234.pdf
\paper.jats
    joss-paper-1234.jats
    image1.png
    image2.png

The paper.jats folder name was problematic when generated in the GitHub action step, where sometimes conflicted with other files present in the submitted sofwtare's repo, so now inara creates the jats output in a folder called jats, it is an internal change. When uploading the files to a branch in the joss-papers repo, there is no problems with the name of the folder so paper.jats is still used there.

[arfon]

Ah great. So there's nothing to do here?

[xuanxu]

Right, I think we're good.