fix(seo): correct sitemap routes

tests/int/sitemap-routes.int.spec.ts

@@ -0,0 +1,210 @@
+import { createRequire } from 'node:module'
+
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
+
+import { getSitemapBaseURL, normalizeSitemapBaseURL } from '@/utilities/sitemapBaseURL.cjs'
+
+type SitemapCollection = 'pages' | 'posts' | 'events' | 'preistraeger'
+
+type SitemapDocument = {
+  _status: 'draft' | 'published'
+  slug?: string
+  spaPath?: string
+  updatedAt?: string
+}
+
+type SitemapFindQuery = {
+  collection: SitemapCollection
+  draft?: boolean
+  overrideAccess?: boolean
+  where?: {
+    _status?: {
+      equals?: string
+    }
+  }
+}
+
+const routeState = vi.hoisted(() => ({
+  docs: {
+    events: [],
+    pages: [],
+    posts: [],
+    preistraeger: [],
+  } as Record<SitemapCollection, SitemapDocument[]>,
+  queries: [] as SitemapFindQuery[],
+}))
+
+vi.mock('@payload-config', () => ({ default: {} }))
+
+vi.mock('next/cache', () => ({
+  unstable_cache: (callback: () => unknown) => callback,
+}))
+
+vi.mock('payload', () => ({
+  getPayload: async () => ({
+    find: async (query: SitemapFindQuery) => {
+      routeState.queries.push(query)
+
+      const isPublishedPublicQuery =
+        query.draft === false &&
+        query.overrideAccess === false &&
+        query.where?._status?.equals === 'published'
+
+      return {
+        docs: isPublishedPublicQuery
+          ? routeState.docs[query.collection].filter((document) => document._status === 'published')
+          : routeState.docs[query.collection],
+      }
+    },
+  }),
+}))
+
+import { GET as getPagesSitemap } from '@/app/(frontend)/(sitemaps)/pages-sitemap.xml/route'
+import { GET as getPostsSitemap } from '@/app/(frontend)/(sitemaps)/posts-sitemap.xml/route'
+
+const productionURL = 'https://www.bayerischer-mittelstandspreis.de'
+
+async function sitemapLocations(response: Response): Promise<string[]> {
+  const xml = await response.text()
+  return [...xml.matchAll(/<loc>([^<]+)<\/loc>/g)].map((match) => match[1])
+}
+
+describe('sitemap route generation', () => {
+  beforeEach(() => {
+    process.env.NEXT_PUBLIC_SERVER_URL = `${productionURL}/`
+    process.env.VERCEL_PROJECT_PRODUCTION_URL = ''
+
+    for (const collection of Object.keys(routeState.docs) as SitemapCollection[]) {
+      routeState.docs[collection] = []
+    }
+    routeState.queries = []
+  })
+
+  afterEach(() => {
+    vi.restoreAllMocks()
+  })
+
+  it.each([
+    {
+      label: 'stored homepage path',
+      document: { _status: 'published', slug: 'startseite', spaPath: '/' },
+      expectedPath: '/',
+    },
+    {
+      label: 'home slug fallback',
+      document: { _status: 'published', slug: 'home' },
+      expectedPath: '/',
+    },
+    {
+      label: 'startseite slug fallback',
+      document: { _status: 'published', slug: 'startseite' },
+      expectedPath: '/',
+    },
+    {
+      label: 'normal stored page path',
+      document: { _status: 'published', slug: 'legacy-contact', spaPath: '/kontakt' },
+      expectedPath: '/kontakt',
+    },
+  ] as const)('emits the $label', async ({ document, expectedPath }) => {
+    routeState.docs.pages = [document]
+
+    await expect(sitemapLocations(await getPagesSitemap())).resolves.toEqual([
+      `${productionURL}${expectedPath}`,
+    ])
+  })
+
+  it.each([
+    { collection: 'posts', prefix: '/presse/blog', slug: 'presseartikel' },
+    { collection: 'events', prefix: '/presse/events', slug: 'preisverleihung' },
+    { collection: 'preistraeger', prefix: '/preistraeger', slug: 'muster-gmbh' },
+  ] as const)(
+    'emits $collection detail routes under $prefix',
+    async ({ collection, prefix, slug }) => {
+      routeState.docs[collection] = [
+        {
+          _status: 'published',
+          slug,
+          spaPath: `${prefix}/${slug}`,
+        },
+      ]
+
+      await expect(sitemapLocations(await getPostsSitemap())).resolves.toContain(
+        `${productionURL}${prefix}/${slug}`,
+      )
+    },
+  )
+
+  it('excludes unpublished documents through the established anonymous public query contract', async () => {
+    routeState.docs.pages = [
+      { _status: 'published', slug: 'sichtbar', spaPath: '/sichtbar' },
+      { _status: 'draft', slug: 'entwurf', spaPath: '/entwurf' },
+    ]
+    routeState.docs.posts = [
+      { _status: 'published', slug: 'sichtbar', spaPath: '/presse/blog/sichtbar' },
+      { _status: 'draft', slug: 'entwurf', spaPath: '/presse/blog/entwurf' },
+    ]
+
+    const locations = [
+      ...(await sitemapLocations(await getPagesSitemap())),
+      ...(await sitemapLocations(await getPostsSitemap())),
+    ]
+
+    expect(locations).toContain(`${productionURL}/sichtbar`)
+    expect(locations).toContain(`${productionURL}/presse/blog/sichtbar`)
+    expect(locations.some((location) => location.includes('entwurf'))).toBe(false)
+    expect(routeState.queries).toHaveLength(4)
+    expect(routeState.queries).toEqual(
+      expect.arrayContaining([
+        expect.objectContaining({
+          draft: false,
+          overrideAccess: false,
+          where: { _status: { equals: 'published' } },
+        }),
+      ]),
+    )
+  })
+})
+
+describe('sitemap production base URL', () => {
+  it.each([
+    [
+      'trailing path and slash',
+      'https://www.bayerischer-mittelstandspreis.de/path///',
+      productionURL,
+    ],
+    ['bare deployment hostname', 'bmp-production.vercel.app', 'https://bmp-production.vercel.app'],
+  ])('normalizes a %s', (_label, configuredURL, expectedURL) => {
+    expect(normalizeSitemapBaseURL(configuredURL)).toBe(expectedURL)
+  })
+
+  it('rejects a localhost production origin', () => {
+    expect(() => normalizeSitemapBaseURL('http://localhost:3000', { production: true })).toThrow(
+      'Production sitemap base URL must be a public HTTPS origin.',
+    )
+  })
+
+  it('fails instead of emitting a fallback when no public URL is configured', () => {
+    expect(() => getSitemapBaseURL({ NODE_ENV: 'production' })).toThrow(
+      'Sitemap base URL is not configured.',
+    )
+  })
+
+  it('uses the normalized production origin for the robots sitemap URLs', () => {
+    const require = createRequire(import.meta.url)
+    const configPath = require.resolve('../../next-sitemap.config.cjs')
+
+    delete require.cache[configPath]
+    process.env.NEXT_PUBLIC_SERVER_URL = `${productionURL}/deployment/path/`
+
+    const config = require(configPath) as {
+      robotsTxtOptions: { additionalSitemaps: string[] }
+      siteUrl: string
+    }
+
+    expect(config.siteUrl).toBe(productionURL)
+    expect(config.robotsTxtOptions.additionalSitemaps).toEqual([
+      `${productionURL}/pages-sitemap.xml`,
+      `${productionURL}/posts-sitemap.xml`,
+    ])
+  })
+})