HTML Galley Production Tool for Janeway

[justingonder]

We need to do an initial exploration of implementing an HTML typesetting tool for our Janeway journals, per UCB accessibility officer.

Janeway has a PANDOC plugin that takes Word files and converts them to HTML. Want to explore whether its viable for our use cases.

[justingonder]

Talk with Jason Colman about what Michigan is doing.

[everreau]

@justingonder can I get some example doc files?

[everreau]

Here's an example with the generic example file: https://pub-jschol2-stg.escholarship.org/uc/item/ttz69zwz

The plugin doesn't work with .doc files only docx. Please let me know if there are other files you want to test.

[justingonder]

AK, CR & JG to look at Glossa and compare outputs to what they are getting from their contractor.

[everreau]

During the typesetting step click "Options" in the far right column of the docx file under "Files for typesetting". Click "Generate HTML". This will create an HTML galley under "Current Galleys" with associated image files.

[justingonder]

Charlotte & Amanda will test a variety of Word docs. Will reconvene after JG is back.

[justingonder]

@everreau Here is a test document we took from Glossa. It's a good start, but the margins are too wide and the images are awfully big. Both things we can attend to with CSS maybe? We tried adding the CSS file you had created for Charlotte, but that didn't help. Neither did adding https://submit.escholarship.org/jschol_static/gloss.css to the Journal Units section of the Django admin.
https://pub-jschol2-stg.escholarship.org/uc/item/ttx2jfnz

[justingonder]

Also noticed that with the Pandoc plugin installed, editors lose the 'convert to PDF' function in the Typsetting stage. Ideally the editors would be able to generate both so that we can provide a PDF download.

[everreau]

@justingonder I don't see any "convert to PDF" option in stage or prd (at least not in the demo journal). Can you describe where you are seeing this?

[everreau]

I moved the link to the css file to the body of the html. The header is not preserved in the indexing process. The regular images look better now. But the orcid icons are still bad. I'll figure out what we need to add here and also if we can add a css file by default. Or maybe we could just add it to the main jschol css.

[everreau]

I see this code in the plugin:

# Undo pandoc guess of height/width attributes of images.
del img["style"]

I'm guessing this is causing the problem. But removing might cause other problems.

[everreau]

This bit is deleting the style but there is also a bit in janeway core that removes css from any html file saved. I'm not sure the best way to deal with this problem. The orcid icon images are actually large images that are resized in the word document. Pandoc just adds a style element with the width and height. We could potentially extract that information and convert it to css but this would have to apply to all the images, it could get messy.

[everreau]

actually. I had a thought maybe I should just resize the images on import. I'm going to give it a try.

[everreau]

I've implemented the image resizing on dev now. The only real issue is that some of the images don't carry DPI information with them so we're using a default dpi. I found different answers to what the default is in word. I tried a couple different and this test document works best with default=72. I don't know how universal this is. We could make it a plugin, press or journal level setting pretty easily but it might work better as something that can be overriden at the time of generation.