create-ailk 0.4.0 → 0.6.0

package/templates/apps/api/src/routes/content/__tests__/search.test.ts

@@ -0,0 +1,450 @@
+/**
+ * @file search.test.ts
+ * GET /v1/content/search — keyword search across published pages (#5831).
+ *
+ * Uses the real content adapter backed by a per-test temp dir (Issue #507).
+ * No internal module mocks — the route exercises the real adapter stack.
+ *
+ * The `category`-tagged fixtures are written here rather than through
+ * `writeFixturePage`, whose frontmatter parameter is `Record<string, string>`
+ * and so cannot express a YAML sequence.
+ */
+
+import * as fs from "fs/promises";
+import * as path from "path";
+
+import type { FastifyInstance } from "fastify";
+
+import { setupContentFixture } from "../../../__tests__/factories/content-fixture";
+import { getContentAdapter } from "../../../lib/content-adapter";
+import { contentRoutes } from "../index";
+import { CONTENT_RATE_LIMIT_UNAUTH, CONTENT_RATE_LIMIT_WINDOW } from "../rate-limit";
+
+const fixture = setupContentFixture();
+
+/** Writes an MDX page with a `categories:` sequence into the locale path. */
+async function writeCategorisedPage(
+  slug: string,
+  title: string,
+  description: string,
+  categories: readonly string[],
+  body: string,
+): Promise<void> {
+  const dir = path.join(fixture.getContentRoot(), "en", "Product");
+  await fs.mkdir(dir, { recursive: true });
+  const frontmatter = [
+    `type: "Product"`,
+    `slug: "${slug}"`,
+    `title: "${title}"`,
+    `description: "${description}"`,
+    ...(categories.length > 0
+      ? ["categories:", ...categories.map((c) => `  - ${c}`)]
+      : []),
+  ].join("\n");
+  await fs.writeFile(
+    path.join(dir, `${slug}.mdx`),
+    `---\n${frontmatter}\n---\n\n${body}\n`,
+    "utf8",
+  );
+}
+
+async function seedRoofPages(): Promise<void> {
+  await writeCategorisedPage(
+    "roof-replacement",
+    "Roof Replacement",
+    "A full roof replacement service.",
+    ["roofing"],
+    "We replace a roof in a day.",
+  );
+  await writeCategorisedPage(
+    "attic-ventilation",
+    "Attic Ventilation",
+    "Ventilation that protects the roof deck.",
+    ["hvac"],
+    "Attic airflow keeps a roof healthy.",
+  );
+  await writeCategorisedPage(
+    "untagged-roof-guide",
+    "Roof Care Guide",
+    "How to look after a roof.",
+    [],
+    "A roof lasts longer when it is inspected.",
+  );
+}
+
+describe("GET /v1/content/search", () => {
+  let app: FastifyInstance;
+
+  beforeEach(async () => {
+    app = require("fastify")();
+    await app.register(contentRoutes);
+  });
+
+  afterEach(async () => {
+    await app.close();
+  });
+
+  it("returns 200 with a JSON-LD ItemList envelope", async () => {
+    await seedRoofPages();
+
+    const res = await app.inject({
+      method: "GET",
+      url: "/v1/content/search?q=roof",
+    });
+
+    expect(res.statusCode).toBe(200);
+    const body = JSON.parse(res.body);
+    expect(body["@context"]).toBe("https://schema.org");
+    expect(body["@type"]).toBe("ItemList");
+    expect(body.query).toBe("roof");
+    expect(Array.isArray(body.items)).toBe(true);
+    expect(body.totalItems).toBe(body.items.length);
+    expect(body.items.map((i: { slug: string }) => i.slug).sort()).toEqual([
+      "attic-ventilation",
+      "roof-replacement",
+      "untagged-roof-guide",
+    ]);
+  });
+
+  it("filters on category — only the tagged page comes back", async () => {
+    await seedRoofPages();
+
+    const res = await app.inject({
+      method: "GET",
+      url: "/v1/content/search?q=roof&category=roofing",
+    });
+
+    expect(res.statusCode).toBe(200);
+    const body = JSON.parse(res.body);
+    expect(body.items.map((i: { slug: string }) => i.slug)).toEqual([
+      "roof-replacement",
+    ]);
+    expect(body.totalItems).toBe(1);
+  });
+
+  it("returns an empty ItemList for a registry slug nothing is tagged with", async () => {
+    await seedRoofPages();
+
+    const res = await app.inject({
+      method: "GET",
+      url: "/v1/content/search?q=roof&category=legal",
+    });
+
+    expect(res.statusCode).toBe(200);
+    const body = JSON.parse(res.body);
+    expect(body.items).toEqual([]);
+    expect(body.totalItems).toBe(0);
+  });
+
+  it("composes the type and category filters", async () => {
+    await seedRoofPages();
+
+    const res = await app.inject({
+      method: "GET",
+      url: "/v1/content/search?q=roof&type=HomePage&category=roofing",
+    });
+
+    expect(res.statusCode).toBe(200);
+    expect(JSON.parse(res.body).items).toEqual([]);
+  });
+
+  it("honours limit", async () => {
+    await seedRoofPages();
+
+    const res = await app.inject({
+      method: "GET",
+      url: "/v1/content/search?q=roof&limit=1",
+    });
+
+    expect(res.statusCode).toBe(200);
+    expect(JSON.parse(res.body).items).toHaveLength(1);
+  });
+
+  it("is not shadowed by the /v1/content/:type route", async () => {
+    const res = await app.inject({
+      method: "GET",
+      url: "/v1/content/search?q=roof",
+    });
+    // A `:type` match would 400 on the type enum; a search match 200s.
+    expect(res.statusCode).toBe(200);
+    expect(JSON.parse(res.body).query).toBe("roof");
+  });
+
+  describe("400s on an invalid query", () => {
+    it("rejects a missing q", async () => {
+      const res = await app.inject({
+        method: "GET",
+        url: "/v1/content/search",
+      });
+      expect(res.statusCode).toBe(400);
+      const body = JSON.parse(res.body);
+      expect(body.error).toBe("Invalid query parameters");
+      expect(Array.isArray(body.details)).toBe(true);
+    });
+
+    it("rejects an empty q", async () => {
+      const res = await app.inject({
+        method: "GET",
+        url: "/v1/content/search?q=",
+      });
+      expect(res.statusCode).toBe(400);
+    });
+
+    it("rejects a category outside the registry", async () => {
+      const res = await app.inject({
+        method: "GET",
+        url: "/v1/content/search?q=roof&category=not-a-real-category",
+      });
+      expect(res.statusCode).toBe(400);
+      const body = JSON.parse(res.body);
+      expect(body.details.some((d: { path: string[] }) => d.path.includes("category"))).toBe(true);
+    });
+
+    it("rejects a type outside the schema.org enum", async () => {
+      const res = await app.inject({
+        method: "GET",
+        url: "/v1/content/search?q=roof&type=NotAType",
+      });
+      expect(res.statusCode).toBe(400);
+    });
+
+    it("rejects a q longer than the 200-char ceiling", async () => {
+      const res = await app.inject({
+        method: "GET",
+        url: `/v1/content/search?q=${"a".repeat(201)}`,
+      });
+      expect(res.statusCode).toBe(400);
+      const body = JSON.parse(res.body);
+      expect(
+        body.details.some((d: { path: string[] }) => d.path.includes("q")),
+      ).toBe(true);
+    });
+
+    it("accepts a q of exactly 200 chars — the same ceiling the MCP tool enforces", async () => {
+      const res = await app.inject({
+        method: "GET",
+        url: `/v1/content/search?q=${"a".repeat(200)}`,
+      });
+      expect(res.statusCode).toBe(200);
+    });
+
+    it("rejects a limit above the 50 ceiling", async () => {
+      const res = await app.inject({
+        method: "GET",
+        url: "/v1/content/search?q=roof&limit=51",
+      });
+      expect(res.statusCode).toBe(400);
+    });
+  });
+
+  /**
+   * D6 — API ↔ MCP parity evidence for #5831.
+   *
+   * The tool and the route must return the same result set for the same
+   * arguments. `apps/api` cannot import `apps/mcp` (the hard import boundary
+   * in the root CLAUDE.md), so the shared seam is asserted instead: the route
+   * must return EXACTLY what `adapter.searchPages` returns for the options the
+   * MCP handler builds from the same arguments. `ailkSearchContent`'s own
+   * suite (`apps/mcp/__tests__/search_content.test.ts`) pins the other half —
+   * tool results ≡ `adapter.searchPages` for the same arguments — so the two
+   * halves compose into the 1:1 claim without either app importing the other.
+   */
+  describe("parity with the adapter the MCP tool calls (D6)", () => {
+    it("deep-equals adapter.searchPages items for the same arguments", async () => {
+      await seedRoofPages();
+
+      const res = await app.inject({
+        method: "GET",
+        url: "/v1/content/search?q=roof&category=roofing&limit=10&locale=en",
+      });
+      expect(res.statusCode).toBe(200);
+
+      // The exact option object `ailkSearchContent` builds for these arguments.
+      const expected = await getContentAdapter().searchPages("roof", {
+        limit: 10,
+        category: "roofing",
+        locale: "en",
+      });
+
+      expect(JSON.parse(res.body).items).toEqual(
+        JSON.parse(JSON.stringify(expected)),
+      );
+    });
+
+    it("deep-equals with no filters at all", async () => {
+      await seedRoofPages();
+
+      const res = await app.inject({
+        method: "GET",
+        url: "/v1/content/search?q=roof",
+      });
+
+      const expected = await getContentAdapter().searchPages("roof", {
+        limit: 10,
+        locale: "en",
+      });
+
+      expect(JSON.parse(res.body).items).toEqual(
+        JSON.parse(JSON.stringify(expected)),
+      );
+    });
+  });
+
+  /**
+   * `server.ts` registers the limiter GLOBALLY with the content ceiling, so
+   * this route is covered without registering one of its own. Asserting the
+   * shared constants is the honest check: a route test that registered its own
+   * limiter would prove nothing about the server.
+   */
+  describe("content rate limit", () => {
+    it("is governed by the same constants as the other content reads", async () => {
+      expect(CONTENT_RATE_LIMIT_UNAUTH).toBe(60);
+      expect(CONTENT_RATE_LIMIT_WINDOW).toBe("1 minute");
+    });
+
+    it("429s past the allowance under a limiter built from those constants", async () => {
+      const limited: FastifyInstance = require("fastify")();
+      const rateLimit = require("@fastify/rate-limit");
+      await limited.register(rateLimit, {
+        max: CONTENT_RATE_LIMIT_UNAUTH,
+        timeWindow: CONTENT_RATE_LIMIT_WINDOW,
+        keyGenerator: () => "shared-test-ip",
+      });
+      await limited.register(contentRoutes);
+
+      for (let i = 0; i < CONTENT_RATE_LIMIT_UNAUTH; i++) {
+        const ok = await limited.inject({
+          method: "GET",
+          url: "/v1/content/search?q=roof",
+        });
+        expect(ok.statusCode).toBe(200);
+      }
+
+      const overLimit = await limited.inject({
+        method: "GET",
+        url: "/v1/content/search?q=roof",
+      });
+      expect(overLimit.statusCode).toBe(429);
+
+      await limited.close();
+    });
+  });
+});
+
+// ─── #5852 / #5910 — the database path: which workspace a search read serves ─
+//
+// Appended, so every MDX case above is untouched. Database-free: the
+// module-level Prisma singleton is faked via the repo's manual-mock
+// convention (see get-page.test.ts's identical block — #5852), and
+// `jest.mock` is hoisted above the imports by ts-jest regardless of where it
+// is written. The MDX cases above never load `lib/prisma.js` — the
+// resolver's import is deferred behind the `CONTENT_ADAPTER=database`
+// branch — so the mock is inert for them.
+
+// eslint-disable-next-line no-restricted-syntax -- lib/prisma.ts wraps the external DB boundary (PrismaClient requires a live connection); __mocks__/prisma.ts is the repo-standard test double for this boundary.
+jest.mock("../../../lib/prisma.js");
+
+import {
+  DEFAULT_ORG_ID,
+  DEFAULT_WORKSPACE_ID,
+} from "@working-theory/database/tenancy";
+
+import { prisma } from "../../../lib/prisma.js";
+
+/** A one-workspace deployment seeded with the well-known default (mirrors get-page.test.ts). */
+function seedOneWorkspaceDeployment(): void {
+  (prisma.workspace.count as jest.Mock).mockResolvedValue(1);
+  (prisma.workspace.findUnique as jest.Mock).mockResolvedValue({
+    id: DEFAULT_WORKSPACE_ID,
+    organizationId: DEFAULT_ORG_ID,
+  });
+}
+
+describe("GET /v1/content/search under CONTENT_ADAPTER=database (#5852 / #5910)", () => {
+  let app: FastifyInstance;
+
+  beforeEach(async () => {
+    jest.clearAllMocks();
+    process.env.CONTENT_ADAPTER = "database";
+    app = require("fastify")();
+    await app.register(contentRoutes);
+  });
+
+  afterEach(async () => {
+    delete process.env.CONTENT_ADAPTER;
+    await app.close();
+  });
+
+  // AC-1 — a resolvable single-workspace deployment returns the search
+  // ItemList, with no workspace id anywhere in the URL, and the underlying
+  // artifact read is bound to the resolved workspace.
+  it("returns the search ItemList for the seeded default workspace, with no workspace id in the URL", async () => {
+    seedOneWorkspaceDeployment();
+    (prisma.artifact.findMany as jest.Mock).mockResolvedValue([]);
+    (prisma.artifactVersion.findMany as jest.Mock).mockResolvedValue([]);
+
+    const url = "/v1/content/search?q=roof";
+    const res = await app.inject({ method: "GET", url });
+
+    expect(res.statusCode).toBe(200);
+    const body = JSON.parse(res.body);
+    expect(body["@type"]).toBe("ItemList");
+    expect(body.items).toEqual([]);
+    // The tenant is resolved server-side; it is never addressable from the wire.
+    expect(url).not.toMatch(/wksp_/);
+    expect(prisma.artifact.findMany).toHaveBeenCalledWith(
+      expect.objectContaining({
+        where: expect.objectContaining({ workspaceId: DEFAULT_WORKSPACE_ID }),
+      }),
+    );
+  });
+
+  // AC-1 — zero workspaces: 404 before any artifact query.
+  it("404s with zero workspaces, before any artifact query", async () => {
+    (prisma.workspace.count as jest.Mock).mockResolvedValue(0);
+
+    const res = await app.inject({
+      method: "GET",
+      url: "/v1/content/search?q=roof",
+    });
+
+    expect(res.statusCode).toBe(404);
+    expect(prisma.artifact.findMany).not.toHaveBeenCalled();
+  });
+
+  // AC-1 — two-or-more workspaces: ambiguous, so no site resolves; 404 before
+  // any artifact query.
+  it("404s with two-or-more workspaces, before any artifact query", async () => {
+    (prisma.workspace.count as jest.Mock).mockResolvedValue(2);
+
+    const res = await app.inject({
+      method: "GET",
+      url: "/v1/content/search?q=roof",
+    });
+
+    expect(res.statusCode).toBe(404);
+    expect(prisma.artifact.findMany).not.toHaveBeenCalled();
+  });
+
+  // The two unresolved arms answer with the SAME 404 body (AC-WCE-05 parity
+  // with get-page/list-pages) — a resolution failure carries no detail an
+  // unauthenticated caller could use to distinguish it from another cause.
+  it("answers both ambiguity arms with identical 404 bodies", async () => {
+    (prisma.workspace.count as jest.Mock).mockResolvedValue(0);
+    const empty = await app.inject({
+      method: "GET",
+      url: "/v1/content/search?q=roof",
+    });
+
+    jest.clearAllMocks();
+    (prisma.workspace.count as jest.Mock).mockResolvedValue(2);
+    const ambiguous = await app.inject({
+      method: "GET",
+      url: "/v1/content/search?q=roof",
+    });
+
+    expect(empty.statusCode).toBe(404);
+    expect(ambiguous.statusCode).toBe(404);
+    expect(ambiguous.body).toBe(empty.body);
+  });
+});

package/templates/apps/api/src/routes/content/get-page.ts

@@ -5,34 +5,69 @@
  * document root, plus the raw frontmatter under `page` for clients that need
  * the unrendered shape.
  *
- * 404 on missing slug, 400 on unknown type.
+ * 404 on missing slug, 400 on unknown type. A deployment whose site cannot be
+ * resolved (#5852) answers with the SAME 404 body as a missing slug.
  *
  * Phase 3 S3: optional `?locale=<locale>` query parameter. Validated against
  * `readLocales()` allowlist; invalid or unknown locale silently falls back to
  * the project's default locale (AC-i18n-07). The adapter applies the fallback
  * chain (locale → default-locale → legacy non-prefixed layout).
+ *
+ * #5890: optional `?include=body` adds the raw published MDX body beside
+ * `page`. It is opt-in: without it the handler takes the path it took before
+ * the parameter existed. An adapter with no `getPageSource` refuses the
+ * parameter with a 400 naming it, rather than 500ing on a missing method.
+ *
+ * #5927: on the `?include=body` path, `page` and `body` come from ONE
+ * `getPageSource` call — never a `getPageBySlug` call plus a second,
+ * independent `getPageSource` call. Two independent adapter reads can
+ * straddle a publish that lands between them (frontmatter from revision N,
+ * body from revision N+1); one read cannot.
  */
-import { SCHEMA_ORG_TYPES, type SchemaOrgType } from "@working-theory/content-adapters";
+import {
+  SCHEMA_ORG_TYPES,
+  type ContentAdapter,
+  type ContentPage,
+  type MdxContentAdapter,
+  type MdxPageSource,
+  type SchemaOrgType,
+} from "@working-theory/content-adapters";
 import { buildJsonLd, type ContentPageInput } from "@working-theory/schema";
 import type { FastifyPluginAsync } from "fastify";
 import { z } from "zod";
 
-import { getContentAdapter } from "../../lib/content-adapter.js";
 import { resolveLocale } from "../../lib/locales.js";
+import { resolvePublicContentAdapter } from "../../services/public-site-resolver.js";
+
+import { SLUG_PATTERN } from "./slug.js";
 
 const paramsSchema = z.object({
   type: z.enum(SCHEMA_ORG_TYPES),
-  slug: z.string().min(1),
+  slug: z.string().regex(SLUG_PATTERN),
 });
 
 const querySchema = z.object({
   locale: z.string().optional(),
+  include: z.enum(["body"]).optional(),
 });
 
+/**
+ * The raw-body capability, probed by the exact name the MDX adapter declares
+ * and the database adapter borrows (#5874) — the same idiom, and the same
+ * name, as `apps/mcp/src/lib/content-adapter.ts`.
+ */
+function hasPageSource(
+  adapter: ContentAdapter,
+): adapter is ContentAdapter & Pick<MdxContentAdapter, "getPageSource"> {
+  return (
+    typeof (adapter as Partial<MdxContentAdapter>).getPageSource === "function"
+  );
+}
+
 export const getPageRoute: FastifyPluginAsync = async (fastify) => {
   fastify.get<{
     Params: { type: string; slug: string };
-    Querystring: { locale?: string };
+    Querystring: { locale?: string; include?: string };
   }>(
     "/v1/content/:type/:slug",
     async (request, reply) => {
@@ -53,15 +88,58 @@ export const getPageRoute: FastifyPluginAsync = async (fastify) => {
       }
 
       const { type, slug } = parsed.data;
+      const { include } = queryParsed.data;
       const locale = resolveLocale(queryParsed.data.locale);
 
       try {
-        const adapter = getContentAdapter();
-        const page = await adapter.getPageBySlug(
-          type as SchemaOrgType,
-          slug,
-          locale,
-        );
+        // Which site this read serves is resolved BEFORE any content query
+        // (#5852). A refusal reuses the SAME not-found body a missing page
+        // gets, below: a distinct body or code would tell an unauthenticated
+        // caller that its slug reached tenancy, which is the enumeration
+        // signal AC-WCE-05 denies.
+        const adapter = await resolvePublicContentAdapter(request);
+        if (adapter === null) {
+          return reply.status(404).send({
+            error: "Page not found",
+            type,
+            slug,
+          });
+        }
+
+        // Refused before any content query, so a caller asking for a body this
+        // deployment cannot serve is never charged a read whose result is then
+        // discarded.
+        let readSource: MdxContentAdapter["getPageSource"] | undefined;
+        if (include === "body") {
+          if (!hasPageSource(adapter)) {
+            return reply.status(400).send({
+              error: "Invalid query parameters",
+              details: [
+                {
+                  path: ["include"],
+                  message: "body is not available from this content adapter",
+                },
+              ],
+            });
+          }
+          readSource = adapter.getPageSource.bind(adapter);
+        }
+
+        // #5927: ONE adapter call supplies everything the response sends.
+        // `readSource` (bound above, only when `include=body`) returns BOTH
+        // the frontmatter and the body from the SAME read, so the two halves
+        // can never straddle a publish that lands between two independent
+        // calls. Without the parameter, `getPageBySlug` is the only call, as
+        // it always was. `source` stays `undefined` on that path — the
+        // `source && {...}` below never attaches a body to the response.
+        let source: MdxPageSource | null | undefined;
+        let page: ContentPage | null;
+        if (readSource) {
+          source = await readSource(type as SchemaOrgType, slug, locale);
+          page = source === null ? null : source.frontmatter;
+        } else {
+          page = await adapter.getPageBySlug(type as SchemaOrgType, slug, locale);
+        }
         if (!page) {
           return reply.status(404).send({
             error: "Page not found",
@@ -88,7 +166,11 @@ export const getPageRoute: FastifyPluginAsync = async (fastify) => {
           ...(page.sections !== undefined && { sections: page.sections }),
         };
         const jsonLd = buildJsonLd(input);
-        return reply.send({ ...jsonLd, page });
+        return reply.send({
+          ...jsonLd,
+          page,
+          ...(source && { body: source.body }),
+        });
       } catch (error) {
         request.log.error({ err: error }, "[Content] getPage failed");
         return reply.status(500).send({ error: "Internal server error" });