Document how to produce a "json feed" containing all content as single JSON file

[trauts2]

📚 Documentation

Similar to Hugo's index.json. I would like to know how to make a JSON output representing all site content and metadata. Like this:

https://www.screencast.com/t/a3spUYaU1J0

This supports a variety of downstream transformations of site content, such as building a custom Algolia index, transforming to other output formats, etc.

Could someone document or suggest how to do this?

Thanks!

Stu

Have you read the Contributing Guidelines on issues?

yes

[yangshun]

We don't have a good way to do that at the moment as we don't aggregate the data model during the bootstrap process into one single data structure, unlike Gatsby. Contents and data fetched within each plugin is scoped to that plugin. The only overall site-wide data we have is the routes, which is used by docusaurus-plugin-sitemap to create sitemap.xml.

But what you have proposed is a step in the right direction. In future, we would try to have similar APIs to build data models in plugins which can be processed downstream in subsequent plugins.

[slorber]

The contentLoaded attribute now accept a "allContent" attribute that permits to access the content loaded by all the plugins.

async contentLoaded({actions: {createData, addRoute}, allContent}) {

Take a look at the official debug plugin for inspiration, as it access the data of all the plugins: https://docusaurus.io/__docusaurus/debug/content

There's also a plan to be able to extend existing content plugins, in case you want to do something with the loaded data of a single plugin. Check #4138, issue is a bit messy though, and there's no official API for this yet but you can work around that.

[kesor]

Migrated from v2 to v3 ... and there is no longer allContent available :(

[slorber]

@kesor this is something we never documented because I wasn't sure it was a good idea.

We later refactored it to another lifecycle method you can try allContentLoaded. Again, still undocumented, we might not make it a public API surface and may need to refactor it if we find a better API design. But the "feature" should stay because we need it ourselves for the debug plugin: https://docusaurus.io/__docusaurus/debug/content

More context here: #9931