To make it easier to read schemas on https://schema.wikimedia.org/#!/, look into tools that can render a schema in a way that is friendlier for humans to read than the raw file without manually duplicating the information in the schema.

Option 1: YAML -> JSON -> wiki page

One option would be to create a bot that would, on a regular cadence:

1. Fetch the latest schema from schema.wikimedia.org
2. Convert from YAML to JSON (https://pyyaml.org)
3. Publish to a wiki page (pywikibot)

Here's an example of the MP web base schema rendered as a JSON page on Wikitech. This option has the advantage of being easily implementable using existing tools, but it does require the maintenance of the bot long term.

See also:

• catalog-to-wiki: A now-deprecated bot that published content from an external source to a wiki page

Option 2: Integrate a tool with schema.wikimedia.org

I looked into several open-source tools that render JSON schemas in Markdown or HTML. However, since it seems that the code for schema.wikimedia.org does not live in a repository, implementing one of these tools would be complicated and likely require engineering resources. This would likely make schema.wikimedia.org more difficult to maintain as well.

See also:

• https://github.com/coveooss/json-schema-for-humans
• https://github.com/adobe/jsonschema2md

Option 3: Custom script and static site

The Blubber docs use a custom script to generate a Markdown file based on a JSON schema and publish the result to a static site. However, this approach is the most complex of the three options.

See also:

• https://gitlab.wikimedia.org/repos/releng/blubber/-/blob/main/util/markdownschema/generate.go

Recommendation

Of these options, I think option 1 would be the easiest to implement, but it's up to Data Products whether the improvement in readability between the raw YAML and the wikitable is worth having to maintain the bot.