@porulle/adapter-pg-search

SearchAdapter powered by PostgreSQL full-text search (tsvector + tsquery). Zero extra infrastructure — your existing Postgres is the search engine.

Usage

import { defineConfig } from "@porulle/core";
import { pgSearchAdapter } from "@porulle/adapter-pg-search";

export default defineConfig({
  search: {
    adapter: pgSearchAdapter({
      query: (sql, params) => yourPgClient.query(sql, params),
      tableName: "search_index",     // default
      dictionary: "english",          // default; pick "simple" for non-English heavy stores
    }),
  },
  // …
});

The query function should be a thin wrapper around your PG client (postgres, pg, etc.) that returns { rows: ... }. The adapter sends parameterized queries — never string-concatenates user input.

When to pick this over Meilisearch

| You want | Use | |---|---| | Zero new services, "just works on the existing DB" | @porulle/adapter-pg-search | | Sub-50ms typo-tolerant search at 10k+ products | @porulle/adapter-meilisearch | | Faceted search with rich aggregations | @porulle/adapter-meilisearch | | <100 products and no search-quality requirements | this adapter is fine |

Notes

• Identifiers (tableName, column names) are validated against ^[a-zA-Z_][a-zA-Z0-9_]*$ to defend against SQL identifier injection. Don't pass user input as tableName.
• The dictionary defaults to english. For multi-locale stores, pick simple and handle stemming at the query layer.

See also

• PostgreSQL Full Text Search
• @porulle/adapter-meilisearch — when you outgrow PG FTS