knowledge-repo
knowledge-repo copied to clipboard
airbnb
Add support for HTML Knowledge Posts
Auto-reviewers: @NiharikaRay @matthewwardrop @earthmancash @danfrankj
We have quite a few HTML posts that we'd need to include in our Knowledge Repo, were we to launch one (which I'm interested to do). Since HTML is valid Markdown, I believe this could be done simply; we'd just need a way to emit the markdown document header.
If this sounds OK, I think I could tackle it. There's already a HTMLConverter class (and it's registered!); I believe this would just involve:
- Documenting a method of embedding header metadata in HTML
- Adding a simple
from_fileclass toHTMLConverter - Adding dependencies (e.g.
beautifulsoup) used within thefrom_filemethod.
Thoughts?
Another option for metadata would be to allow for specification on the command-line, when add is specified. The path (-p) option is already custom to that command, so I guess it wouldn't be too hard to add options for each required header metadata value.
Could be limited to html, or be propagated to other conversion types as well.
Just an idea :)
@bcipolli Thanks for reaching out about this.
Regarding HTML -> Markdown, have you looked at html2text or other similar libraries that already do this? Adding headers in an automated way will require some work. Would you want the headers to be in the HTML source (in meta tags or some such); or in a comment at the top of the document; or would you want to pass them in at runtime?
@matthewwardrop I would use META tags, and would be happy to allow passing in at runtime as well... if that sounds good.
Would definitely use a library to insert the HTML, would not roll my own :)
@bcipolli We'd love to have your contributions! Feel free to submit patches! If possible, perhaps write the HTML support in a separate patch from the support for run-time specification of missing tags. The runtime specification of missing tags probably belongs as part of the retooling effort in #308 , so once the HTML patch is ready, drop a line and we can discuss the best way to extend the tooling!
@merkliopas We prototyped this, but didn't get to a place where we felt confident to push a PR. At this point, we've moved to using Blogdown / Hugo via R - I'm ok closing this out.
It would be nice to support HTML format so that we can avoid rendering Rmarkdown in the run time, which might require installing additional packages for rendering in the docker container/host.
Agreed HTML support would be nice. I'll leave this issue open in case anyone decides to contribute with a patch.
