Algolia Search
Integrate Algolia Search with Fumadocs
If you're using Algolia's free tier, you have to display their logo on your search dialog.
Indexing
The default behavior creates a record for each paragraph in your document which officially recommended by Algolia.
Each record contains searchable attributes title
, section
, and content
.
section
is optional, it only exists in paragraphs under headings. Headings and
paragraphs are all indexed as a record, and the distinct attribute is
page_id
.
Setup
Install algoliasearch
Sign up and obtain the required API keys for your search. Store these credentials in environment variables.
Create and sync your indexes with the sync
function.
The indexes
variable should be provided by your content source provider. If you are using Fumadocs MDX, see Search Index Generation.
Notably, it requires a structured
property, this can be obtained from structurize, or sometimes supplied by the content source you are using.
You may make it a script and manually sync with node ./update-index.mjs
, or
integrate it with your CI/CD pipeline.
Notice that it expects the url
property of a page to be unique. In other words, you shouldn't have two pages with the same
url.
If you are running the script with
TSX or other similar Typescript
executors, ensure to name it .mts
for best ESM compatibility.
To search documents on the client side, you can use the Fumadocs UI Search Dialog, or bring your own implementation.
In addition, the useAlgoliaSearch
hook can handle state management with SWR.
Customise Attributes & Settings
While the default attributes might not suit your case, you can pass extra_data
to index options for adding extra fields to each record.
To customize the default index settings, set index settings, and update
documents with updateDocuments(...)
separately.
Last updated on