@qbobjx/plugins

Official OBJX plugins, including timestamps, snake case naming, soft delete, tenant scope, and audit trail.

Install

npm install @qbobjx/plugins

Quick Usage

import { col, defineModel } from '@qbobjx/core';
import {
  createSnakeCaseNamingPlugin,
  createSoftDeletePlugin,
  createTenantScopePlugin,
} from '@qbobjx/plugins';

export const Account = defineModel({
  table: 'accounts',
  columns: {
    id: col.int().primary(),
    tenantId: col.text(),
    deletedAt: col.timestamp().nullable(),
  },
  plugins: [
    createSnakeCaseNamingPlugin(),
    createTenantScopePlugin(),
    createSoftDeletePlugin(),
  ],
});

Included Plugins

• createTimestampsPlugin()
• createSnakeCaseNamingPlugin()
• createSoftDeletePlugin()
• createTenantScopePlugin()
• createAuditTrailPlugin()

Snake Case Naming

Use createSnakeCaseNamingPlugin() when your model keys are camelCase but your physical database columns are snake_case.

import { col, defineModel } from '@qbobjx/core';
import { createSnakeCaseNamingPlugin } from '@qbobjx/plugins';

export const Account = defineModel({
  table: 'accounts',
  columns: {
    id: col.int().primary(),
    tenantId: col.text(),
    createdAt: col.timestamp(),
  },
  plugins: [
    createSnakeCaseNamingPlugin({
      exclude: ['id'],
      overrides: {
        createdAt: 'created_on',
      },
    }),
  ],
});

The compiler emits physical names like tenant_id, while hydrated rows continue to use model keys like tenantId.

Important: this plugin should be attached at model definition time. It updates column metadata in the onModelDefine hook, so adding it only as a session-global plugin is too late for remapping.

Repository examples using this pattern:

• examples/complex-runtime
• examples/express-api
• examples/nestjs-api

Tenant Scope Vs PostgreSQL RLS

createTenantScopePlugin() injects tenant predicates into OBJX-generated SQL. It does not replace PostgreSQL RLS policies that depend on set_config(...) / current_setting(...).

If your database enforces tenant isolation through PostgreSQL RLS, configure createPostgresSession({ executionContextSettings: ... }) in @qbobjx/postgres-driver and execute protected work inside session.transaction(...).

PostgreSQL Runtime Suite (experimental)

OBJX now ships a PostgreSQL-focused runtime suite directly in @qbobjx/plugins with internal defaults intended for objx_internal tables managed by the ORM.

import {
  createPostgresEventsPlugin,
  createPostgresPreset,
  createPostgresQueuePlugin,
  createPostgresSearchPlugin,
} from '@qbobjx/plugins';

const runtimePreset = createPostgresPreset({
  schema: 'objx_internal',
  queue: { defaultQueue: 'default' },
  events: { notifyChannel: 'objx_events' },
});

// habilitar apenas queue (cron/jobs)
const onlyQueue = createPostgresPreset({
  include: ['queue'],
  queue: { defaultQueue: 'cron' },
});

// habilitar apenas events
const onlyEvents = createPostgresPreset({
  include: ['events'],
  events: { notifyChannel: 'events_only' },
});

const queuePlugin = createPostgresQueuePlugin({
  schema: 'objx_internal',
  autoProvision: true,
  defaultQueue: 'critical',
});

const eventsPlugin = createPostgresEventsPlugin({
  schema: 'objx_internal',
  autoProvision: true,
});

const searchPlugin = createPostgresSearchPlugin({
  defaultLanguage: 'english',
  rankFunction: 'ts_rank_cd',
});

Included PostgreSQL runtime plugin factories:

• createPostgresSearchPlugin()
• createPostgresQueuePlugin()
• createPostgresEventsPlugin()
• createPostgresCachePlugin()
• createPostgresVectorPlugin()
• createPostgresTimeseriesPlugin()
• createPostgresJsonPlugin()
• createPostgresSecurityPlugin()
• createPostgresObservabilityPlugin()
• createPostgresPreset()
• createPostgresInternalSchemaSql()

Observação: cada plugin vive em módulo próprio dentro de src/postgres/*; o preset apenas compõe os plugins que você escolher em include.

Runtime API (experimental)

Além dos builders SQL, o módulo PostgreSQL agora expõe um runtime de alto nível para executar queue, events, cache, search, vector, timeseries, json, security e observability.

Você pode usar o runtime com qualquer executor compatível:

import { createPostgresRuntime } from '@qbobjx/plugins';

const runtime = createPostgresRuntime({
  async execute(sql, params = []) {
    return db.execute(sql, params);
  },
});

await runtime.queue.enqueue({
  queueName: 'default',
  jobName: 'sync-project',
  payload: { projectId: 1 },
});

await runtime.events.publish({
  eventName: 'project.created',
  payload: { projectId: 1 },
});

O runtime também inclui:

• provisionInternalSchema() para bootstrap automático de tabelas internas (runtime_migrations, queue, outbox, cache).
• withRequest({ executionContext, transactionId }) para execução com contexto explícito.
• runQueueWorker(...) e runEventDispatcher(...) para loops básicos de worker/dispatcher.
• metrics() com contadores operacionais por domínio.

Driver session integration

O caminho recomendado com o driver oficial é integrar os plugins PostgreSQL com createPostgresSession(...) e criar o runtime a partir da própria sessão.

import { col, createModelRegistry, defineModel } from '@qbobjx/core';
import { createPostgresSession } from '@qbobjx/postgres-driver';
import {
  createPostgresEventsPlugin,
  createPostgresIntegration,
  createPostgresQueuePlugin,
  createPostgresRuntimeFromSession,
} from '@qbobjx/plugins';

const RuntimeConfig = defineModel({
  table: 'runtime_config',
  columns: {
    id: col.int().primary(),
    tenantId: col.text(),
  },
  plugins: [
    createPostgresQueuePlugin({
      schema: 'objx_internal',
      defaultQueue: 'default',
    }),
    createPostgresEventsPlugin({
      schema: 'objx_internal',
      notifyChannel: 'objx_events',
    }),
  ],
});

const registry = createModelRegistry();
registry.register(RuntimeConfig);

const integration = createPostgresIntegration(registry);

const session = createPostgresSession({
  pool,
  executionContextSettings: integration.executionContextSettings,
});

const runtime = createPostgresRuntimeFromSession(session, {
  source: registry,
  config: integration.config,
});

Essa integração garante que:

• o runtime use a configuração real vinda dos plugins
• o driver oficial aplique set_config(...) de forma transacional quando necessário
• fila, outbox, cache e demais APIs compartilhem a mesma base de configuração

PostgreSQL Runtime API guide

A runtime has one required contract: an executor with execute(sql, params, request?).

type Executor = {
  execute<T = unknown>(
    sql: string,
    params?: readonly unknown[],
    request?: { executionContext?: unknown; transactionId?: string },
  ): Promise<T>;
};

Também existe integração pronta com ObjxSession / createPostgresSession(...):

• createPostgresSessionExecutor(session, options?)
• createPostgresRuntimeFromSession(session, options?)
• createPostgresRuntimeFromObjxSession(session, options?)

Bootstrapping and lifecycle

import { createPostgresRuntime } from '@qbobjx/plugins';

const runtime = createPostgresRuntime(executor);

// Creates objx_internal + runtime_migrations + queue/outbox/cache tables.
await runtime.provisionInternalSchema({
  pluginName: 'objx-postgres-runtime',
  version: '1',
});

// Bind an explicit request/transaction context for all subsequent calls.
const contextualRuntime = runtime.withRequest({ executionContext });

Queue API

await runtime.queue.enqueue({
  queueName: 'default',
  jobName: 'invoice.generate',
  payload: { invoiceId: 'inv_1' },
  priority: 10,
  maxAttempts: 8,
});

const job = await runtime.queue.dequeue('worker-a');
if (job) {
  // ...process
  await runtime.queue.complete(1);
}

• enqueue: creates queued job rows.
• dequeue: claims a pending job with SKIP LOCKED semantics.
• complete: marks running job as done.
• fail: applies retry/backoff or dead status depending on attempts.

Events API (Outbox)

await runtime.events.publish({
  eventName: 'project.created',
  payload: { projectId: 'p_1' },
  aggregateId: 'p_1',
  aggregateType: 'project',
  idempotencyKey: 'project.created:p_1',
});

const batch = await runtime.events.dispatchBatch(100);
for (const event of batch as Array<{ id: number }>) {
  // ...deliver to webhook/broker
  await runtime.events.ack(event.id);
}

Cache API

const value = await runtime.cache.getOrCompute(
  'project:summary:p_1',
  async () => ({ id: 'p_1', name: 'API revamp' }),
  { tags: ['project', 'summary'] },
);

• get, set, getOrCompute, pruneExpired, metrics.

Search, Vector, JSON, Security, Observability

await runtime.search.query({ table: 'docs', query: 'postgres runtime', limit: 20 });
await runtime.vector.similarity({ table: 'docs', vector: [0.12, 0.44, 0.05] });
const where = runtime.json.wherePath('metadata', '$.tenant == "acme"');
await runtime.security.setLocalTenant('acme');
const top = await runtime.observability.topStatements(10);

Background handles

const queueHandle = runtime.startQueueWorker(async (job) => {
  // business logic
}, { workerId: 'worker-1' });

const eventHandle = runtime.startEventDispatcher(async (event) => {
  // delivery logic
}, { batchSize: 200 });

// later
queueHandle.stop();
eventHandle.stop();
await queueHandle.done;
await eventHandle.done;

Runtime metrics

const snapshot = runtime.metrics();
console.log(snapshot.queue.enqueued, snapshot.events.published, snapshot.cache.hits);

Functional project example: examples/postgres-runtime.