feat: sitemaps

app/[locale]/sitemap.xml/route.ts

@@ -0,0 +1,69 @@
+import { NextRequest, NextResponse } from "next/server";
+import { XMLBuilder } from "fast-xml-parser";
+import {
+  generateSitemapUrl,
+  getSitemapData,
+  getSiteMapNewsData,
+} from "@/lib/api/sitemap";
+
+export async function GET(
+  request: NextRequest,
+  { params: { locale } }: LocaleProps
+) {
+  const pages = await getSitemapData(locale);
+  const pageData = pages.map(({ uri, dateUpdated }) => {
+    return {
+      loc: generateSitemapUrl(uri, locale),
+      lastmod: dateUpdated,
+    };
+  });
+  const { siteTitle, news } = await getSiteMapNewsData(locale);
+
+  const today = new Date();
+  const recentNewsThreshold = new Date(
+    today.getTime() - 1000 * 60 * 60 * 24 * 2
+  );
+
+  const newsData = news.map(({ uri, dateUpdated, title, date }) => {
+    const entry = {
+      loc: generateSitemapUrl(uri, locale),
+      lastmod: dateUpdated,
+    };
+
+    if (new Date(date) > recentNewsThreshold) {
+      entry["news:news"] = {
+        "news:publication": {
+          "news:name": siteTitle,
+          "news:language": locale,
+        },
+        "news:publication_date": date,
+        "news:title": title,
+      };
+    }
+
+    return entry;
+  });
+
+  const data = {
+    "?xml": {
+      $version: "1.0",
+      $encoding: "UTF-8",
+    },
+    urlset: {
+      $xmlns: "http://www.sitemaps.org/schemas/sitemap/0.9",
+      "$xmlns:news": "http://www.google.com/schemas/sitemap-news/0.9",
+      url: pageData.concat(newsData),
+    },
+  };
+  const builder = new XMLBuilder({
+    attributeNamePrefix: "$",
+    arrayNodeName: "url",
+    ignoreAttributes: false,
+  });
+  const output = builder.build(data);
+
+  return new NextResponse(output, {
+    status: 200,
+    headers: { "Content-Type": "application/xml; charset=utf-8" },
+  });
+}

app/sitemap_index.xml/route.ts

@@ -0,0 +1,30 @@
+import { NextResponse } from "next/server";
+import { XMLBuilder } from "fast-xml-parser";
+import { generateSitemapUrl } from "@/lib/api/sitemap";
+import { languages } from "@/lib/i18n/settings";
+
+export async function GET() {
+  const data = {
+    "?xml": {
+      $version: "1.0",
+      $encoding: "UTF-8",
+    },
+    sitemapindex: {
+      $xmlns: "http://www.sitemaps.org/schemas/sitemap/0.9",
+      sitemap: languages.map((language) => {
+        return { loc: generateSitemapUrl("sitemap.xml", language, true) };
+      }),
+    },
+  };
+  const builder = new XMLBuilder({
+    attributeNamePrefix: "$",
+    arrayNodeName: "sitemap",
+    ignoreAttributes: false,
+  });
+  const output = builder.build(data);
+
+  return new NextResponse(output, {
+    status: 200,
+    headers: { "Content-Type": "application/xml; charset=utf-8" },
+  });
+}

cypress/e2e/api/sitemaps.cy.js

@@ -0,0 +1,55 @@
+import { XMLValidator, XMLParser } from "fast-xml-parser";
+import { languages } from "../../../lib/i18n/settings";
+
+const url = "/sitemap_index.xml";
+
+context("GET /server_index.xml", () => {
+  it("returns a well formed XML file", () => {
+    cy.request({
+      url,
+      method: "GET",
+    }).then(({ body, headers }) => {
+      expect(headers["content-type"]).to.eq("application/xml; charset=utf-8");
+
+      const valid = XMLValidator.validate(body);
+
+      expect(valid).to.eq(true);
+    });
+  });
+  it("contains a sitemap for each locale", () => {
+    cy.request({
+      url,
+      method: "GET",
+    }).then(({ body }) => {
+      const parser = new XMLParser();
+      const {
+        sitemapindex: { sitemap },
+      } = parser.parse(body);
+
+      expect(sitemap.length).to.eq(languages.length);
+    });
+  });
+  it("contains valid sitemap links", () => {
+    cy.request({
+      url,
+      method: "GET",
+    }).then(({ body }) => {
+      const parser = new XMLParser();
+      const {
+        sitemapindex: { sitemap },
+      } = parser.parse(body);
+
+      sitemap.forEach(({ loc }) => {
+        cy.request({ url: loc, method: "GET" }).then(({ body, headers }) => {
+          expect(headers["content-type"]).to.eq(
+            "application/xml; charset=utf-8"
+          );
+
+          const valid = XMLValidator.validate(body);
+
+          expect(valid).to.eq(true);
+        });
+      });
+    });
+  });
+});

lib/api/homepage/index.ts

@@ -7,8 +7,7 @@ import {
 } from "@/lib/api/fragments/content-blocks";
 import { getLinkFields, linkFragment } from "@/lib/api/fragments/link";
 import queryAPI from "@/lib/api/client/query";
-
-const CRAFT_HOMEPAGE_URI = "__home__";
+import { CRAFT_HOMEPAGE_URI } from "@/lib/constants";
 
 export const getHomepage = async (locale: string, previewToken?: string) => {
   const site = getSiteFromLocale(locale);

lib/api/sitemap/index.ts

@@ -0,0 +1,124 @@
+import { gql } from "@urql/core";
+import { getSiteFromLocale } from "@/lib/helpers/site";
+import queryAPI from "@/lib/api/client/query";
+import tags from "@/lib/api/client/tags";
+import { CRAFT_HOMEPAGE_URI } from "@/lib/constants";
+import { fallbackLng } from "@/lib/i18n/settings";
+
+interface PageMetadata {
+  uri: string;
+  dateUpdated: string;
+}
+const baseUrl = process.env.NEXT_PUBLIC_BASE_URL || "";
+
+export const generateSitemapUrl = (
+  uri: string,
+  locale: string,
+  preserveLocale = false
+) => {
+  const segments = uri === CRAFT_HOMEPAGE_URI ? [] : [uri];
+
+  if (preserveLocale || locale !== fallbackLng) {
+    segments.unshift(locale);
+  }
+
+  segments.unshift(baseUrl);
+
+  return segments.join("/");
+};
+
+export const getSiteMapNewsData = async (locale: string) => {
+  const site = getSiteFromLocale(locale);
+
+  const query = gql`
+    query NewsSitemapData($site: [String]) {
+      newsEntries(site: $site) {
+        ... on news_post_Entry {
+          title
+          uri
+          date
+          dateUpdated
+        }
+      }
+      globalSets(site: $site) {
+        ... on siteInfo_GlobalSet {
+          siteTitle
+        }
+      }
+    }
+  `;
+
+  const { data } = await queryAPI({
+    query,
+    variables: { site },
+    fetchOptions: {
+      next: { tags: [tags.globals] },
+    },
+  });
+
+  const { newsEntries, globalSets } = data;
+  const { siteTitle } = globalSets.filter((set) =>
+    Object.hasOwn(set, "siteTitle")
+  )[0];
+
+  return { siteTitle, news: newsEntries };
+};
+
+export const getSitemapData = async (
+  locale: string
+): Promise<Array<PageMetadata>> => {
+  const site = getSiteFromLocale(locale);
+
+  const query = gql`
+    query SitemapData($site: [String]) {
+      pages: entries(
+        site: $site
+        section: ["pages"]
+        type: ["not", "redirectPage"]
+      ) {
+        uri
+        dateUpdated
+      }
+      homepage: homepageEntries(site: $site) {
+        ... on homepage_homepage_Entry {
+          uri
+          dateUpdated
+        }
+      }
+      staff: staffProfilesEntries(site: $site) {
+        ... on staffProfiles_staffProfiles_Entry {
+          dateUpdated
+          uri
+        }
+      }
+      slideshows: slideshowsEntries(site: $site) {
+        ... on slideshows_slideshow_Entry {
+          dateUpdated
+          uri
+        }
+      }
+      glossary: glossaryTermsEntries(site: $site) {
+        ... on glossaryTerms_glossaryTerm_Entry {
+          dateUpdated
+          uri
+        }
+      }
+      voice: slideshowsEntries(site: $site) {
+        ... on slideshows_slideshow_Entry {
+          dateUpdated
+          uri
+        }
+      }
+    }
+  `;
+
+  const { data } = await queryAPI({
+    query,
+    variables: { site },
+    fetchOptions: {
+      next: { tags: [tags.globals] },
+    },
+  });
+
+  return Object.values<any>(data).flat();
+};

lib/constants/index.ts

@@ -0,0 +1 @@
+export const CRAFT_HOMEPAGE_URI = "__home__";

package.json

@@ -67,6 +67,7 @@
     "add": "^2.0.6",
     "classnames": "^2.3.1",
     "convert": "^4.13.0",
+    "fast-xml-parser": "^4.5.0",
     "feed": "^4.2.2",
     "focus-trap": "^7.0.0",
     "focus-visible": "^5.1.0",

yarn.lock

@@ -6695,6 +6695,13 @@ fast-levenshtein@^2.0.6, fast-levenshtein@~2.0.6:
   resolved "https://registry.yarnpkg.com/fast-levenshtein/-/fast-levenshtein-2.0.6.tgz#3d8a5c66883a16a30ca8643e851f19baa7797917"
   integrity sha512-DCXu6Ifhqcks7TZKY3Hxp3y6qphY5SJZmrWMDrKcERSOXWQdMhU9Ig/PYrzyw/ul9jOIyh0N4M0tbC5hodg8dw==
 
+fast-xml-parser@^4.5.0:
+  version "4.5.0"
+  resolved "https://registry.yarnpkg.com/fast-xml-parser/-/fast-xml-parser-4.5.0.tgz#2882b7d01a6825dfdf909638f2de0256351def37"
+  integrity sha512-/PlTQCI96+fZMAOLMZK4CWG1ItCbfZ/0jx7UIJFChPNrx7tcEgerUgWbeieCM9MfHInUDyK8DWYZ+YrywDJuTg==
+  dependencies:
+    strnum "^1.0.5"
+
 fastest-levenshtein@^1.0.16:
   version "1.0.16"
   resolved "https://registry.yarnpkg.com/fastest-levenshtein/-/fastest-levenshtein-1.0.16.tgz#210e61b6ff181de91ea9b3d1b84fdedd47e034e5"
@@ -11613,6 +11620,11 @@ striptags@^3.2.0:
   resolved "https://registry.yarnpkg.com/striptags/-/striptags-3.2.0.tgz#cc74a137db2de8b0b9a370006334161f7dd67052"
   integrity sha512-g45ZOGzHDMe2bdYMdIvdAfCQkCTDMGBazSw1ypMowwGIee7ZQ5dU0rBJ8Jqgl+jAKIv4dbeE1jscZq9wid1Tkw==
 
+strnum@^1.0.5:
+  version "1.0.5"
+  resolved "https://registry.yarnpkg.com/strnum/-/strnum-1.0.5.tgz#5c4e829fe15ad4ff0d20c3db5ac97b73c9b072db"
+  integrity sha512-J8bbNyKKXl5qYcR36TIO8W3mVGVHrmmxsd5PAItGkmyzwJvybiw2IVq5nqd0i4LSNSkB/sx9VHllbfFdr9k1JA==
+
 style-loader@^3.3.1:
   version "3.3.2"
   resolved "https://registry.yarnpkg.com/style-loader/-/style-loader-3.3.2.tgz#eaebca714d9e462c19aa1e3599057bc363924899"