Meilisearch

zero exposes Meilisearch through the unified ctx.Search handle (the same handle used for Solr). Meilisearch is a fast, HTTP-based search engine; zero talks to it with a small pure-Zig client.

Configuration

MEILI_HOST=http://localhost:7700
MEILI_INDEX=movies
MEILI_API_KEY=             # optional; sent as Authorization: Bearer

API

ctx.Search is unchanged — index, fetch, delete, and query documents regardless of backend:

pub fn addDoc(ctx: *Context) !void {
    try ctx.Search.?.index(ctx, "movies",
        "[{\"id\":1,\"title\":\"Inception\"}]");
}

pub fn search(ctx: *Context) !void {
    const res = try ctx.Search.?.query(ctx, "movies", "inception");
    defer ctx.allocator.free(res);
    try ctx.json(res);
}

• index(ctx, collection, doc_json) — upsert a JSON doc.
• query(ctx, collection, q) — search.
• get(ctx, collection, id) — fetch a doc by id (null on 404).
• delete(ctx, collection, id) — delete a doc by id.
• lastError() — surfaces the upstream status.

ctx.Search is optional — guard with if (ctx.Search) |s| { ... }. Free returned slices with ctx.allocator.

Caveats

• The search call sends only {"q":"<term>"}. Filters, facets, pagination, and attributesToRetrieve are not supported yet.
• No circuit breaker is wired yet.

See the runnable zero-search example and the implementation in src/datasource/search/meili.zig.