feat: add dynamic sitemap generation and API endpoint

apps/trustlab/src/lib/data/common/sitemap.js

@@ -0,0 +1,156 @@
+import * as Sentry from "@sentry/nextjs";
+
+import { site } from "@/trustlab/utils";
+
+function normalizePathname(pathname) {
+  if (!pathname || typeof pathname !== "string") {
+    return null;
+  }
+
+  if (pathname === "/") {
+    return pathname;
+  }
+
+  const trimmed = pathname.trim();
+  if (!trimmed) {
+    return null;
+  }
+
+  const withLeadingSlash = trimmed.startsWith("/") ? trimmed : `/${trimmed}`;
+  return withLeadingSlash.replace(/\/+$/, "");
+}
+
+function getAbsoluteUrl(pathname) {
+  const normalizedPathname = normalizePathname(pathname);
+  if (!normalizedPathname) {
+    return null;
+  }
+
+  const siteUrl = site.url.replace(/\/+$/, "");
+  return `${siteUrl}${normalizedPathname}`;
+}
+
+function getLastModified(doc) {
+  const rawDate = doc?.updatedAt || doc?.createdAt;
+  if (!rawDate) {
+    return null;
+  }
+
+  const parsedDate = new Date(rawDate);
+  if (Number.isNaN(parsedDate.getTime())) {
+    return null;
+  }
+
+  return parsedDate.toISOString();
+}
+
+function toSitemapEntry(doc, pathname) {
+  const url = getAbsoluteUrl(pathname);
+  if (!url) {
+    return null;
+  }
+
+  return {
+    url,
+    lastModified: getLastModified(doc),
+  };
+}
+
+function warnOnMissingPathname(collection, doc) {
+  Sentry.logger.warn(`Collection item without \`pathname\` in sitemap`, {
+    collection,
+    slug: doc?.slug,
+  });
+}
+
+async function getPagesEntries(api) {
+  const { docs } = await api.getCollection("pages", {
+    pagination: false,
+    select: {
+      pathname: true,
+      slug: true,
+      parent: true,
+      breadcrumbs: true,
+      updatedAt: true,
+      createdAt: true,
+    },
+    where: {
+      and: [
+        {
+          _status: {
+            equals: "published",
+          },
+        },
+        {
+          slug: {
+            not_in: ["404", "500"],
+          },
+        },
+      ],
+    },
+  });
+
+  return docs
+    .map((doc) => {
+      if (!doc?.pathname) {
+        warnOnMissingPathname("pages", doc);
+        return null;
+      }
+
+      return toSitemapEntry(doc, doc.pathname);
+    })
+    .filter(Boolean);
+}
+
+async function getOpportunitiesEntries(api) {
+  const { docs } = await api.getCollection("opportunities", {
+    pagination: false,
+    select: {
+      pathname: true,
+      slug: true,
+      type: true,
+      updatedAt: true,
+      createdAt: true,
+      date: true,
+    },
+  });
+
+  return docs
+    .map((doc) => {
+      if (!doc?.pathname) {
+        warnOnMissingPathname("opportunities", doc);
+        return null;
+      }
+
+      return toSitemapEntry(doc, doc.pathname);
+    })
+    .filter(Boolean);
+}
+
+async function getSitemapEntries(api) {
+  const [pages, opportunities] = await Promise.all([
+    getPagesEntries(api),
+    getOpportunitiesEntries(api),
+  ]);
+
+  return [...pages, ...opportunities].sort((left, right) =>
+    left.url.localeCompare(right.url),
+  );
+}
+
+async function buildSitemapXml(api) {
+  const entries = await getSitemapEntries(api);
+  const xmlEntries = entries
+    .map(({ url, lastModified }) => {
+      const lastModifiedNode = lastModified
+        ? `\n    <lastmod>${lastModified}</lastmod>`
+        : "";
+
+      return `  <url>\n    <loc>${url}</loc>${lastModifiedNode}\n  </url>`;
+    })
+    .join("\n");
+
+  return `<?xml version="1.0" encoding="UTF-8"?>\n<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n${xmlEntries}\n</urlset>\n`;
+}
+
+export default buildSitemapXml;

apps/trustlab/src/lib/data/index.js

@@ -1,3 +1,8 @@
-export { getPageStaticPaths, getPageStaticProps, getRobotsTxt } from "./local";
+export {
+  getPageStaticPaths,
+  getPageStaticProps,
+  getRobotsTxt,
+  getSitemapXml,
+} from "./local";
 
 export default undefined;

apps/trustlab/src/lib/data/local/index.js

@@ -1,4 +1,5 @@
 import { getPageProps, getPagePaths } from "@/trustlab/lib/data/common";
+import buildSitemapXml from "@/trustlab/lib/data/common/sitemap";
 import api from "@/trustlab/lib/payload";
 
 export async function getPageStaticPaths() {
@@ -21,4 +22,8 @@ export async function getRobotsTxt() {
   return siteSettings?.robotsTxt;
 }
 
+export async function getSitemapXml() {
+  return buildSitemapXml(api);
+}
+
 export default undefined;

apps/trustlab/src/pages/sitemap.xml.page.js

@@ -0,0 +1,30 @@
+import * as Sentry from "@sentry/nextjs";
+
+import { getSitemapXml } from "@/trustlab/lib/data";
+
+export async function getServerSideProps({ res }) {
+  try {
+    const sitemapXml = await getSitemapXml();
+    res.setHeader(
+      "Cache-Control",
+      "public, max-age=3600, stale-while-revalidate=86400",
+    );
+    res.setHeader("Content-Type", "application/xml; charset=utf-8");
+    res.write(sitemapXml);
+    res.end();
+    return {
+      props: {},
+    };
+  } catch (error) {
+    Sentry.captureException(error);
+    res.statusCode = 500;
+    res.end();
+    return {
+      props: {},
+    };
+  }
+}
+
+export default function SitemapXml() {
+  return null;
+}