From 2b56c451778a5af04c480b6e7b02fbea98376308 Mon Sep 17 00:00:00 2001
From: David Cumps <david@cumps.eu>
Date: Wed, 21 Oct 2026 22:09:07 +0000
Subject: [PATCH] fix(seo): rebase only the exact default authority, insert
 SITE_URL literally

Review follow-up: the origin regex also matched a port
(https://www.bentopdf.com:8443) and userinfo
(https://www.bentopdf.com@other.example), which are different origins;
the lookahead now excludes ':' and '@'. The replacement is a callback,
so `$&`/`$1` in SITE_URL stay literal. Tests cover both.
---
 scripts/site-origin.mjs       | 12 ++++++++----
 src/tests/site-origin.test.ts | 15 +++++++++++++++
 2 files changed, 23 insertions(+), 4 deletions(-)

diff --git a/scripts/site-origin.mjs b/scripts/site-origin.mjs
index 920bdef8..ae5e9a5b 100644
--- a/scripts/site-origin.mjs
+++ b/scripts/site-origin.mjs
@@ -4,16 +4,20 @@
  */
 export const DEFAULT_SITE_URL = 'https://www.bentopdf.com';
 
-const DEFAULT_ORIGIN_RE = /https:\/\/www\.bentopdf\.com(?![\w.-])/g;
+// The exact authority only: not a longer host (`.com.example`), a port
+// (`:8443`, a different origin) or userinfo (`@other.example`, another host).
+const DEFAULT_ORIGIN_RE = /https:\/\/www\.bentopdf\.com(?![\w.:@-])/g;
 
 /**
  * Rewrites absolute URLs on the default origin to `siteUrl` (no trailing
  * slash). Used for hand-written pages that carry the official origin
  * literally, so a build with a custom SITE_URL emits canonicals on its own
- * host. Other hosts, including look-alikes such as
- * `https://www.bentopdf.com.example`, are left untouched.
+ * host. Other origins are left untouched: look-alike hosts such as
+ * `https://www.bentopdf.com.example`, other ports and userinfo URLs.
  */
 export function rebaseDefaultOrigin(html, siteUrl) {
   if (siteUrl === DEFAULT_SITE_URL) return html;
-  return html.replace(DEFAULT_ORIGIN_RE, siteUrl);
+  // A callback keeps `siteUrl` literal: a string replacement would expand
+  // `$&`, `$1`, … inside it.
+  return html.replace(DEFAULT_ORIGIN_RE, () => siteUrl);
 }
diff --git a/src/tests/site-origin.test.ts b/src/tests/site-origin.test.ts
index 3ff994f4..16a7e4e7 100644
--- a/src/tests/site-origin.test.ts
+++ b/src/tests/site-origin.test.ts
@@ -35,6 +35,21 @@ describe('rebaseDefaultOrigin', () => {
     expect(rebaseDefaultOrigin(html, 'https://pdf.example.org')).toBe(html);
   });
 
+  it('leaves other ports and userinfo URLs alone', () => {
+    const html =
+      'href="https://www.bentopdf.com:8443/blog/" href="https://www.bentopdf.com@other.example/"';
+    expect(rebaseDefaultOrigin(html, 'https://pdf.example.org')).toBe(html);
+  });
+
+  it('inserts the site URL literally', () => {
+    expect(
+      rebaseDefaultOrigin(
+        'href="https://www.bentopdf.com/blog/"',
+        'https://example.org/$&'
+      )
+    ).toBe('href="https://example.org/$&/blog/"');
+  });
+
   it('is a no-op for the default site URL', () => {
     expect(rebaseDefaultOrigin(PAGE, DEFAULT_SITE_URL)).toBe(PAGE);
   });