astro-d1-search

Site search for Astro, backed by Cloudflare D1 (SQLite FTS5): build-time indexing, an injected search API, bm25 + recency ranking.

Running in production on zander.wtf/search — see theExamples guide for its full source.

What you get

Build-time indexing of any markdown or MDX — content collections, standalone pages, anything on disk — into a D1 full-text index.
An injected API endpoint (GET /api/search) with bm25 + recency ranking, highlighted snippets, type filtering, input validation and edge caching.
A query function (searchIndex) for server-rendered search pages that hit the D1 binding directly, with no HTTP round trip.
Content-type agnostic: types are strings you choose, so one index can serve site-wide search, per-section search, and even several sites at once.

Requires the @astrojs/cloudflare adapter. The rest of your site stays fully static — only the search endpoint runs server-side.

Getting started

Install the package, create a D1 database, and get a working search endpoint.

Indexing content

How sources map files to URLs, which frontmatter is read, and what the markdown stripper keeps.

Search pages

Query D1 directly from a server-rendered Astro page, or hit the injected endpoint from the client.

Deploying

CI ordering, API token scopes, local development, and sharing one database across sites.

Examples

A real site running the package end to end, with links to every file involved.

Configuration

Every option accepted by the integration, with its default and what it affects.

API endpoint

Request parameters, response shape, validation rules, caching and injection safety.

Ranking

How bm25 weights and the recency multiplier combine into a score, and how to tune them.

Exports

The three entry points, every exported function and type, and which one to import where.