Meilisearch

@stndrds/adapter-meilisearch is the recommended search adapter for standards. It indexes records on write and serves typed queries from React hooks.

Setup

Pass a MeilisearchSearchAdapter instance to SupabaseDatabaseAdapter as the search option. The adapter is mandatory — SupabaseDatabaseAdapter requires it at construction time.

// @noverify
import { MeilisearchSearchAdapter } from "@stndrds/adapter-meilisearch";
import { SupabaseDatabaseAdapter } from "@stndrds/adapter-supabase";
import { SchemaModule } from "@stndrds/adapter-nestjs";

const search = new MeilisearchSearchAdapter({
  host: process.env.MEILISEARCH_HOST!,
  apiKey: process.env.MEILISEARCH_API_KEY!,
});

const adapter = new SupabaseDatabaseAdapter(supabase, { search });

// In your NestJS app module:
SchemaModule.forRoot({
  adapter,
  registry,
  search: {
    reconcileOnStartup: true,      // compare PG vs Meili on boot
    reconcileIntervalMs: 900_000,  // periodic safety net (15 min)
    fullReindexThreshold: 0.2,     // reindex when delta > 20%
  },
})

The search key in SchemaModule.forRoot accepts a SearchConfig with three optional fields: reconcileOnStartup (default true), reconcileIntervalMs (default 15 minutes), and fullReindexThreshold (default 20%). All three have sensible defaults — the search key itself is optional.

Env vars

• MEILISEARCH_HOST — Meilisearch URL (e.g. http://localhost:7700)
• MEILISEARCH_API_KEY — admin key for indexing and settings updates

Sync triggers

The adapter hooks into the database adapter's CRUD events through the SchemaSearchModule registered by SchemaModule. Records sync to Meilisearch automatically on create, update, and delete — no manual reindex is needed for routine changes.

The sync path uses an event bus internally, so indexing is decoupled from the HTTP response. A write returns to the caller as soon as PostgreSQL commits; Meilisearch is updated asynchronously.

Manual reindex

@stndrds/adapter-meilisearch exports reindexAllRecords for bulk reindexing. Use it when setting up Meilisearch on an existing dataset, after a major schema change, or when reconcile reports a large delta.

// @noverify
import { reindexAllRecords } from "@stndrds/adapter-meilisearch";
import { withTenantContext, asTenantId } from "@stndrds/schema";

await withTenantContext(asTenantId("your-tenant-id"), async () => {
  const result = await reindexAllRecords(adapter, searchIndexer, {
    batchSize: 500,
    onProgress: (done, total) => console.log(`${done}/${total}`),
  });
  console.log(`Indexed ${result.indexed} records in ${result.duration}ms`);
});

reindexAllRecords requires a tenant context. In single-tenant setups, use the defaultTenantId configured in SchemaModule. The batchSize option (default 500) controls both the PostgreSQL query page size and the Meilisearch bulk-add chunk size.

A reconcile helper is also exported for differential sync — it compares PG and Meilisearch counts and only reindexes divergent records, which is faster than a full reindex on large datasets.