Migrate to archive.sparkpost.com with host-conditional noindex

components/site/seo.tsx

@@ -17,7 +17,7 @@ const SEO = (props: SeoProps): JSX.Element => {
       <meta name="viewport" content="width=device-width, initial-scale=1.0" />
       <link rel="canonical" href={router.asPath}></link>
       <link rel="shortcut icon" type="image/png" href="/favicon.png" />
-      <meta name="robots" content="index, follow, max-image-preview:large, max-snippet:-1, max-video-preview:-1" />
+      <meta name="robots" content="noindex, nofollow" />
       {title && (
         <>
           <title>{title}</title>

netlify.toml

@@ -9,6 +9,13 @@ package = "@netlify/plugin-nextjs"
   for = "/*"
 
   [headers.values]
+    # Applies to every host this deploy serves. The host-conditional-noindex
+    # edge function strips this header (and the noindex <meta> tag) for
+    # support.sparkpost.com requests during the CloudFront migration gap, so
+    # only archive.sparkpost.com (and deploy previews) actually receive it.
+    # See netlify/edge-functions/host-conditional-noindex.ts.
+    X-Robots-Tag = "noindex, nofollow"
+
     Strict-Transport-Security = '''
     max-age=31536000;
     includeSubDomains;'''
@@ -47,7 +54,7 @@ package = "@netlify/plugin-nextjs"
       'self'
       data:
       fonts.gstatic.com
-      support.sparkpost.com;
+      archive.sparkpost.com;
     connect-src
       *
       data:

netlify/edge-functions/host-conditional-noindex.ts

@@ -0,0 +1,99 @@
+// Temporary edge function for the support.sparkpost.com → archive.sparkpost.com
+// migration. Until the CloudFront redirect distribution is live (then DNS for
+// support.sparkpost.com flips off Netlify), this single Netlify deploy serves
+// both hostnames. The desired SEO behavior is asymmetric:
+//
+//   archive.sparkpost.com: served as built — noindex <meta>, X-Robots-Tag, and
+//     Disallow: / in robots.txt all intact, so search engines drop the archive.
+//   support.sparkpost.com: keep search engines indexing as before, so existing
+//     link equity is preserved until the CloudFront layer can 301 it to bird.com.
+//
+// Without this function, the noindex signals (set in seo.tsx, netlify.toml, and
+// next-sitemap.js) would apply to both hostnames equally — which would start
+// deindexing support.sparkpost.com before the 301s exist, losing link equity
+// that would otherwise transfer to bird.com via the CloudFront redirects.
+//
+// This function differentiates by Host header:
+//   - support.sparkpost.com: override X-Robots-Tag with "all", strip the noindex
+//     <meta> from HTML responses, and override /robots.txt with a permissive
+//     version.
+//   - archive.sparkpost.com or any other host (deploy previews, *.netlify.app):
+//     pass through unmodified.
+//
+// REMOVE THIS FUNCTION (and the netlify.toml registration) once the CloudFront
+// cutover is complete — support.sparkpost.com will no longer hit Netlify, so the
+// hostname-conditional logic becomes dead code.
+
+import type { Context } from '@netlify/edge-functions';
+
+const INDEX_HOST = 'support.sparkpost.com';
+
+// Match <meta name="robots" content="...noindex..."> in either attribute order
+// and with whitespace variations. Only the noindex/nofollow robots meta is
+// targeted — any other meta tags (description, og:*, viewport, etc.) are left
+// alone.
+const NOINDEX_META_PATTERNS: RegExp[] = [
+  /<meta\s[^>]*name=["']robots["'][^>]*content=["'][^"']*noindex[^"']*["'][^>]*\/?>/gi,
+  /<meta\s[^>]*content=["'][^"']*noindex[^"']*["'][^>]*name=["']robots["'][^>]*\/?>/gi,
+];
+
+export default async (request: Request, context: Context): Promise<Response | void> => {
+  const host = request.headers.get('host') ?? '';
+
+  // Only act on support.sparkpost.com. Every other host — including
+  // archive.sparkpost.com, *.netlify.app deploy URLs, and deploy previews —
+  // passes through with whatever the build emits (i.e. noindex stays on).
+  if (host !== INDEX_HOST) return;
+
+  const url = new URL(request.url);
+
+  // Override /robots.txt with a permissive version so Google can crawl
+  // support.sparkpost.com normally. Without this, the built robots.txt's
+  // `Disallow: /` would block Google from re-crawling, which would prevent it
+  // from seeing the 301s once CloudFront comes online.
+  if (url.pathname === '/robots.txt') {
+    return new Response('User-agent: *\n', {
+      status: 200,
+      headers: { 'Content-Type': 'text/plain; charset=utf-8' },
+    });
+  }
+
+  // For everything else, fetch the response normally, then mutate it.
+  const response = await context.next();
+
+  // Override the X-Robots-Tag set in netlify.toml. `delete()` on this header is
+  // silently ignored — Netlify re-applies the netlify.toml [[headers]] block
+  // after the edge function returns, but it respects values WE set. So we
+  // overwrite with "all" (canonical "ignore any prior noindex; index normally"),
+  // which Google treats as equivalent to no header at all.
+  response.headers.set('X-Robots-Tag', 'all');
+
+  // Only HTML responses can carry the noindex <meta> tag. Skip non-HTML
+  // (images, JS, CSS, JSON) to avoid pointlessly buffering large bodies.
+  const contentType = response.headers.get('content-type') ?? '';
+  if (!contentType.includes('text/html')) {
+    return response;
+  }
+
+  let body = await response.text();
+  for (const pattern of NOINDEX_META_PATTERNS) {
+    body = body.replace(pattern, '');
+  }
+
+  // Rebuild the response with the mutated body. content-length is dropped so
+  // Netlify recomputes it. content-encoding is dropped because response.text()
+  // already decoded any gzip/brotli the origin sent — leaving the header on
+  // would tell the client to decompress a now-plain-text body and fail.
+  const headers = new Headers(response.headers);
+  headers.delete('content-length');
+  headers.delete('content-encoding');
+  return new Response(body, {
+    status: response.status,
+    statusText: response.statusText,
+    headers,
+  });
+};
+
+export const config = {
+  path: '/*',
+};

next-sitemap.js

@@ -1,11 +1,12 @@
 module.exports = {
-  siteUrl: process.env.SITE_URL || 'https://support.sparkpost.com',
+  siteUrl: process.env.SITE_URL || 'https://archive.sparkpost.com',
   generateRobotsTxt: true,
   sitemapSize: 5000,
   robotsTxtOptions: {
     policies: [
       {
         userAgent: '*',
+        disallow: ['/'],
       },
     ],
   },

tsconfig.json

@@ -24,5 +24,5 @@
     "@context/*": ["context/*"]
   },
   "include": ["next-env.d.ts", "yaml.d.ts", "**/*.ts", "**/*.tsx"],
-  "exclude": ["node_modules"]
+  "exclude": ["node_modules", "netlify"]
 }