Pav 74 repensar geracao keywords scraper

.env.example

@@ -51,6 +51,8 @@ CACHE_TTL_MS=600000
 REDIS_KEY_PREFIX=vagas-full
 KEYWORDS_REDIS_KEY=vagas-full:keywords
 KEYWORDS_STORAGE_MODE=env
+# Mantém GET /keywords ativo, mas bloqueia POST /keywords e publicação no kwsync.
+KWSYNC_ENABLED=false
 
 # Legacy flags kept for compatibility
 HEADLESS=false

LOCAL_DEVELOPMENT.md

@@ -234,13 +234,13 @@ Como criar usuário para testes:
 
 ## 7.1 Subir stack completa (recomendado para onboarding)
 
-1. Criar rede:
+1; Criar rede:
 
 ```bash
 docker network create vagas-net
 ```
 
-2. Subir infra + app + migrate:
+2; Subir infra + app + migrate:
 
 ```bash
 docker compose -f docker-compose.infra.yml -f docker-compose.yml -f docker-compose.migrate.yml up --build -d
@@ -298,10 +298,10 @@ Use o comando da seção de Docker.
 
 Portas esperadas:
 
-- frontend: http://localhost:5173
-- front_admin: http://localhost:5174
-- backend: http://localhost:3001
-- scraper-go: http://localhost:8081
+- frontend: <http://localhost:5173>
+- front_admin: <http://localhost:5174>
+- backend: <http://localhost:3001>
+- scraper-go: <http://localhost:8081>
 
 ## Caminho B: Node local (frontend + backend)
 
@@ -345,27 +345,27 @@ Observação importante para o Caminho B:
 
 URLs principais:
 
-- App principal: http://localhost:5173
-- Login: http://localhost:5173/login
-- Cadastro: http://localhost:5173/register
+- App principal: <http://localhost:5173>
+- Login: <http://localhost:5173/login>
+- Cadastro: <http://localhost:5173/register>
 - Dashboard app: /home, /dashboard, /vagas, /mentoria, /perfil, /ajuda
 - Callback OAuth: /auth/callback
 
 Backend:
 
-- Health: http://localhost:3001/health
-- Swagger: http://localhost:3001/docs
-- Metrics: http://localhost:3001/metrics
+- Health: <http://localhost:3001/health>
+- Swagger: <http://localhost:3001/docs>
+- Metrics: <http://localhost:3001/metrics>
 
 Scraper:
 
-- Health: http://localhost:8081/health
-- Metrics: http://localhost:8081/metrics
-- Admin jobs count: http://localhost:8081/admin/jobs/count
+- Health: <http://localhost:8081/health>
+- Metrics: <http://localhost:8081/metrics>
+- Admin jobs count: <http://localhost:8081/admin/jobs/count>
 
 Front admin:
 
-- http://localhost:5174
+- <http://localhost:5174>
 - rota de login: /login
 - rotas principais: /dashboard, /users, /scrapers, /observability, /audit, /permissions, /settings
 
@@ -430,7 +430,7 @@ Abaixo, os testes manuais sugeridos para os módulos principais.
 
 Passos:
 
-1. Acesse http://localhost:5173/login
+1. Acesse <http://localhost:5173/login>
 2. Tente enviar vazio
 3. Informe credenciais inválidas
 4. Informe credenciais válidas
@@ -445,7 +445,7 @@ Resultado esperado:
 
 Passos:
 
-1. Acesse http://localhost:5173/register
+1. Acesse <http://localhost:5173/register>
 2. Preencha campos obrigatórios
 3. Teste telefone opcional vazio
 4. Teste telefone válido
@@ -500,7 +500,7 @@ Resultado esperado:
 
 Passos:
 
-1. Acesse http://localhost:5174/login
+1. Acesse <http://localhost:5174/login>
 2. Faça login com conta com permissão
 3. Navegue por dashboard/users/scrapers/observability/audit/permissions/settings
 

backend/.env.example

@@ -38,6 +38,8 @@ CORS_ALLOWED_ORIGINS=http://localhost:5173,http://localhost:5174
 DATABASE_URL=postgresql://vagas:vagas@localhost:5432/vagas
 VALKEY_URL=redis://localhost:6379/0
 CACHE_TTL_MS=600000
+# Mantém GET /keywords ativo, mas bloqueia POST /keywords e publicação no kwsync.
+KWSYNC_ENABLED=false
 
 # Scraping behavior
 WAIT_BETWEEN_SEARCHES_MS=5000

backend/src/config.ts

@@ -22,6 +22,7 @@ export interface AppConfig {
   emailFromAddress: string;
   emailFromName: string;
   emailQueueAttempts: number;
+  kwsyncEnabled: boolean;
 }
 
 function parseBoolean(value: string | undefined, fallback: boolean): boolean {
@@ -75,5 +76,6 @@ export function getConfig(): AppConfig {
     emailFromAddress: process.env.EMAIL_FROM_ADDRESS?.trim() ?? "",
     emailFromName: process.env.EMAIL_FROM_NAME?.trim() ?? "",
     emailQueueAttempts: parseNumber(process.env.EMAIL_QUEUE_ATTEMPTS, 3),
+    kwsyncEnabled: parseBoolean(process.env.KWSYNC_ENABLED, false),
   };
 }

backend/src/lib/kwsync.ts

@@ -1,4 +1,5 @@
 import type { RedisClientType } from "redis";
+import { getConfig } from "../config";
 import { logger } from "../logger";
 
 // Chave absoluta global. O Go lerá exatamente esse namespace.
@@ -21,6 +22,14 @@ export async function publish(
   source: KeywordEvent["source"] = "user",
   userId?: string,
 ): Promise<void> {
+  if (!getConfig().kwsyncEnabled) {
+    logger.info(
+      { keyword, source, userId },
+      "kwsync: publicação ignorada porque KWSYNC_ENABLED=false",
+    );
+    return;
+  }
+
   const event: KeywordEvent = {
     keyword,
     source,
@@ -53,6 +62,13 @@ export async function publishBatch(
   source: KeywordEvent["source"] = "user",
 ): Promise<void> {
   if (keywords.length === 0) return;
+  if (!getConfig().kwsyncEnabled) {
+    logger.info(
+      { count: keywords.length, source },
+      "kwsync: lote ignorado porque KWSYNC_ENABLED=false",
+    );
+    return;
+  }
 
   const now = new Date().toISOString();
   const payloads = keywords.map((keyword) =>

backend/src/routes/keywords.routes.ts

@@ -4,6 +4,7 @@ import { keywords } from "../db/schema";
 import { ownedBy } from "../lib/authorization/ownership";
 import { getCache } from "../lib/cache";
 import { publish } from "../lib/kwsync";
+import { getConfig } from "../config";
 
 export const keywordsRoutes = Router();
 
@@ -55,13 +56,22 @@ keywordsRoutes.get("/", async (req, res) => {
  *     responses:
  *       202:
  *         description: Keyword enfileirada — o Go decide se persiste
+ *       403:
+ *         description: Submissão de keywords por usuário desabilitada
  *       400:
  *         description: Dados inválidos
  */
 keywordsRoutes.post("/", async (req, res) => {
   const userId = req.session?.userId;
   if (!userId) return res.status(401).json({ message: "Não autenticado." });
 
+  if (!getConfig().kwsyncEnabled) {
+    return res.status(403).json({
+      ok: false,
+      message: "Submissão de keywords por usuário está desabilitada.",
+    });
+  }
+
   const raw = req.body?.keyword;
   const keyword = typeof raw === "string" ? raw.trim() : "";
 

backend/tests/unit/app.test.ts

@@ -122,6 +122,7 @@ const DEFAULT_PAGINATED = (ids: string[]) => ({
 describe("jobsApiApp", () => {
   beforeEach(() => {
     vi.clearAllMocks();
+    process.env.KWSYNC_ENABLED = "false";
 
     mocks.parsePagination.mockReturnValue(DEFAULT_PAGINATION);
     mocks.paginate.mockImplementation((ids: string[]) =>
@@ -709,6 +710,8 @@ describe("jobsApiApp", () => {
   });
 
   it("POST /keywords enfileira keyword e retorna 202", async () => {
+    process.env.KWSYNC_ENABLED = "true";
+
     const app = createJobsApiApp();
     const res = await request(app)
       .post("/keywords")
@@ -733,7 +736,26 @@ describe("jobsApiApp", () => {
     });
   });
 
+  it("POST /keywords retorna 403 quando kwsync está desabilitado", async () => {
+    const app = createJobsApiApp();
+
+    const res = await request(app)
+      .post("/keywords")
+      .send({ keyword: "Rust" })
+      .expect(403);
+
+    expect(res.body).toEqual({
+      ok: false,
+      message: "Submissão de keywords por usuário está desabilitada.",
+    });
+    expect(mocks.dbInsert).not.toHaveBeenCalled();
+    expect(mocks.getCache).not.toHaveBeenCalled();
+    expect(mocks.publish).not.toHaveBeenCalled();
+  });
+
   it("POST /keywords retorna 400 quando keyword está ausente", async () => {
+    process.env.KWSYNC_ENABLED = "true";
+
     const app = createJobsApiApp();
 
     const res = await request(app).post("/keywords").send({}).expect(400);
@@ -745,6 +767,8 @@ describe("jobsApiApp", () => {
   });
 
   it("POST /keywords retorna 400 quando keyword é string vazia", async () => {
+    process.env.KWSYNC_ENABLED = "true";
+
     const app = createJobsApiApp();
 
     const res = await request(app)
@@ -758,6 +782,8 @@ describe("jobsApiApp", () => {
   });
 
   it("POST /keywords retorna 400 quando keyword não é string", async () => {
+    process.env.KWSYNC_ENABLED = "true";
+
     // Arrays não são strings — a rota rejeita com 400 se keyword.trim() não existir
     // ou com 500 se explodir antes. Ajusta a expectativa ao comportamento real da rota:
     // req.body?.keyword?.trim() em um array retorna undefined → cai no if → 400

backend/tests/unit/libs/kwsync.test.ts

@@ -28,7 +28,20 @@ function getCall(client: ReturnType<typeof makeClient>, index = 0) {
 }
 
 describe("publish", () => {
-  beforeEach(() => vi.clearAllMocks());
+  beforeEach(() => {
+    vi.clearAllMocks();
+    process.env.KWSYNC_ENABLED = "true";
+  });
+
+  it("ignora publicação quando KWSYNC_ENABLED=false", async () => {
+    process.env.KWSYNC_ENABLED = "false";
+
+    const client = makeClient();
+    await publish(client as any, "React");
+
+    expect(client.lPush).not.toHaveBeenCalled();
+    expect(mocks.loggerInfo).toHaveBeenCalledOnce();
+  });
 
   it("chama lPush com a chave correta", async () => {
     const client = makeClient();
@@ -79,14 +92,27 @@ describe("publish", () => {
 });
 
 describe("publishBatch", () => {
-  beforeEach(() => vi.clearAllMocks());
+  beforeEach(() => {
+    vi.clearAllMocks();
+    process.env.KWSYNC_ENABLED = "true";
+  });
 
   it("nao chama lPush quando keywords e array vazio", async () => {
     const client = makeClient();
     await publishBatch(client as any, []);
     expect(client.lPush).not.toHaveBeenCalled();
   });
 
+  it("ignora lote quando KWSYNC_ENABLED=false", async () => {
+    process.env.KWSYNC_ENABLED = "false";
+
+    const client = makeClient();
+    await publishBatch(client as any, ["Java", "Node.js"]);
+
+    expect(client.lPush).not.toHaveBeenCalled();
+    expect(mocks.loggerInfo).toHaveBeenCalledOnce();
+  });
+
   it("chama lPush uma unica vez com array de payloads", async () => {
     const client = makeClient();
     await publishBatch(client as any, ["Java", "Node.js", "Go"]);

frontend/tests/unit/new_dashboard/jobs.test.tsx

@@ -4,6 +4,7 @@ import { describe, expect, it, vi } from "vitest";
 import { AddJobModal } from "@/domains/new_dashboard/components/jobs/AddJobModal";
 import { JobDetailModal } from "@/domains/new_dashboard/components/jobs/JobDetailModal";
 import { JobFilter } from "@/domains/new_dashboard/components/jobs/JobFilter";
+import { FormattedJobDescription } from "@/domains/new_dashboard/components/jobs/FormattedJobDescription";
 import { JobRow } from "@/domains/new_dashboard/components/jobs/JobRow";
 import { JobTab } from "@/domains/new_dashboard/components/jobs/JobTab";
 import { JobTable } from "@/domains/new_dashboard/components/jobs/JobTable";
@@ -340,6 +341,55 @@ describe("new_dashboard job components", () => {
     expect(screen.queryByText(/<h3>/i)).not.toBeInTheDocument();
   });
 
+  it("renderiza texto puro, tags semânticas e links inseguros da descrição", () => {
+    const plainRender = render(
+      <FormattedJobDescription description={"Linha 1\nLinha 2"} />,
+    );
+
+    expect(screen.getByText(/linha 1/i)).toBeInTheDocument();
+    expect(screen.getByText(/linha 2/i)).toBeInTheDocument();
+    expect(plainRender.container.querySelector("p")).toHaveClass(
+      "whitespace-pre-wrap",
+    );
+
+    plainRender.unmount();
+
+    const { container } = render(
+      <FormattedJobDescription
+        description={[
+          "<h1>Título principal</h1>",
+          "<h2>Subtítulo</h2>",
+          "<h4>Grupo</h4>",
+          "<blockquote>Citação</blockquote>",
+          "<pre><code>npm test</code></pre>",
+          "<hr>",
+          '<a href="javascript:alert(1)">Link bloqueado</a>',
+          '<a href="/vaga">Link relativo seguro</a>',
+          "<img src=x onerror=alert(1)>",
+          "<div><span>Conteúdo preservado</span></div>",
+        ].join("")}
+      />,
+    );
+
+    expect(
+      screen.getByRole("heading", { name: "Título principal", level: 1 }),
+    ).toBeInTheDocument();
+    expect(
+      screen.getByRole("heading", { name: "Subtítulo", level: 2 }),
+    ).toBeInTheDocument();
+    expect(
+      screen.getByRole("heading", { name: "Grupo", level: 4 }),
+    ).toBeInTheDocument();
+    expect(container.querySelector("blockquote")).toHaveTextContent("Citação");
+    expect(container.querySelector("pre")).toHaveTextContent("npm test");
+    expect(container.querySelector("hr")).toBeInTheDocument();
+    expect(screen.getByText("Link bloqueado").tagName).toBe("SPAN");
+    expect(screen.getByRole("link", { name: "Link relativo seguro" }))
+      .toHaveAttribute("href", "http://localhost:3000/vaga");
+    expect(container.querySelector("img")).not.toBeInTheDocument();
+    expect(screen.getByText("Conteúdo preservado")).toBeInTheDocument();
+  });
+
   it("valida e salva uma vaga manual nova", () => {
     const onAddJob = vi.fn();
     const onClose = vi.fn();