From 611d9da6588d6796059526572a911da4b2bb03f0 Mon Sep 17 00:00:00 2001
From: Sarthak Agrawal <sarthakagrawal927@gmail.com>
Date: Sat, 15 Aug 2026 04:54:40 +0530
Subject: [PATCH] test(seo): add sitemap/canonical parity regression check

Closes #47

Issue #47: the live sitemap listed /faq and /login with the homepage
canonical. The runtime fixes (self-canonical for /faq, /login removed
from the sitemap, noindex on /login) shipped in #48. This adds the
missing regression guard so the parity cannot silently regress:

- For every URL in public/sitemap.xml, resolve its landing-astro source
  page and assert it declares a self-canonical (canonicalPath == its own
  pathname). The homepage keeps the Layout default '/'.
- Assert /login never re-enters the sitemap.
- Reuses the existing sitemapUrls() helper and AGENT_SURFACE catalog.
---
 src/agent-surfaces.test.ts | 47 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 47 insertions(+)

diff --git a/src/agent-surfaces.test.ts b/src/agent-surfaces.test.ts
index 59487a4..3b73e43 100644
--- a/src/agent-surfaces.test.ts
+++ b/src/agent-surfaces.test.ts
@@ -4,6 +4,7 @@ import { describe, expect, it } from 'vitest';
 import { AGENT_SURFACE } from './agent-edge.mjs';
 
 const readPublic = (path: string) => readFileSync(resolve(process.cwd(), 'public', path), 'utf8');
+const readRepo = (path: string) => readFileSync(resolve(process.cwd(), path), 'utf8');
 
 function sitemapUrls(): string[] {
   return [...readPublic('sitemap.xml').matchAll(/<loc>([^<]+)<\/loc>/g)].map(
@@ -11,6 +12,25 @@ function sitemapUrls(): string[] {
   );
 }
 
+/**
+ * Map a sitemap pathname to the landing-astro source page that renders it.
+ * Astro uses `index.astro` for directory routes and `<name>.astro` for leaf
+ * routes; the homepage is `pages/index.astro`.
+ */
+function landingSourceForPath(pathname: string): string {
+  if (pathname === '/') return 'landing-astro/src/pages/index.astro';
+  const base = pathname.replace(/^\//, '');
+  // Prefer a leaf route (e.g. /faq -> pages/faq.astro); fall back to a
+  // directory route (e.g. /changelog -> pages/changelog/index.astro).
+  const leaf = `landing-astro/src/pages/${base}.astro`;
+  try {
+    readFileSync(resolve(process.cwd(), leaf));
+    return leaf;
+  } catch {
+    return `landing-astro/src/pages/${base}/index.astro`;
+  }
+}
+
 describe('public agent surface parity', () => {
   it('keeps the sitemap limited to canonical HTML surfaces', () => {
     const catalogUrls = AGENT_SURFACE.catalog.surfaces
@@ -80,4 +100,31 @@ describe('public agent surface parity', () => {
     expect(AGENT_SURFACE.llmsFullTxt).toBe(readPublic('llms-full.txt'));
     expect(AGENT_SURFACE.catalog.surfaces).toEqual(JSON.parse(readPublic('api-ai.json')).surfaces);
   });
+
+  // Regression for #47: every sitemap URL must render a self-canonical
+  // (canonical == its own URL). Account routes like /login must stay out of
+  // the sitemap and must not be assigned the homepage canonical.
+  it('every sitemap URL declares a self-canonical matching its own path', () => {
+    const urls = sitemapUrls();
+    expect(urls).not.toContain('https://read.significanthobbies.com/login');
+
+    for (const url of urls) {
+      const pathname = new URL(url).pathname;
+      const sourcePath = landingSourceForPath(pathname);
+      const source = readRepo(sourcePath);
+
+      // The Layout canonical defaults to '/', so the homepage needs no
+      // explicit declaration. Every other sitemap route must declare an
+      // explicit canonicalPath equal to its own pathname.
+      if (pathname === '/') {
+        // Home must not override the default '/' canonical with a different path.
+        const override = source.match(/canonicalPath\s*=\s*"([^"]+)"/);
+        expect(override ? override[1] : '/').toBe('/');
+      } else {
+        const match = source.match(/canonicalPath\s*=\s*"([^"]+)"/);
+        expect(match, `${sourcePath} must declare a self-canonical for ${pathname}`).not.toBeNull();
+        expect(match?.[1]).toBe(pathname);
+      }
+    }
+  });
 });