From 17a7cf97a922d6579b3c112a66ef21779e2aa6b8 Mon Sep 17 00:00:00 2001
From: Aaron Moore <aaron.moore@bbc.co.uk>
Date: Tue, 22 Oct 2024 18:40:35 +0100
Subject: [PATCH] Use URL class for adding lite extension more accurately

---
 .../litePageTransforms/index.test.ts          |  9 +++--
 .../litePageTransforms/transformAnchorTags.ts | 37 ++++++++++++-------
 2 files changed, 28 insertions(+), 18 deletions(-)

diff --git a/src/server/Document/Renderers/litePageTransforms/index.test.ts b/src/server/Document/Renderers/litePageTransforms/index.test.ts
index b97cc83fc3e..669ad44c3e2 100644
--- a/src/server/Document/Renderers/litePageTransforms/index.test.ts
+++ b/src/server/Document/Renderers/litePageTransforms/index.test.ts
@@ -1,24 +1,25 @@
 import litePageTransforms from '.';
 
 describe('litePageTransforms', () => {
+  beforeAll(() => {
+    process.env.SIMORGH_BASE_URL = 'https://www.bbc.com';
+  });
   describe('anchor tags', () => {
     it('should append .lite suffix to valid hrefs', () => {
       const html = `
-        <a href="https://www.bbc.com/news">News</a>
         <a href="https://www.bbc.com/serbian/lat">News</a>
         <a href="https://www.bbc.com/mundo">News</a>
         <a href="https://www.bbcrussian.com/news">News</a>
-        <a href="/news">News</a>
+        <a href="https://www.bbc.com/news">News</a>
       `;
 
       const modifiedHtml = litePageTransforms(html);
 
       expect(modifiedHtml).toEqual(`
-        <a href="https://www.bbc.com/news.lite">News</a>
         <a href="https://www.bbc.com/serbian/lat.lite">News</a>
         <a href="https://www.bbc.com/mundo.lite">News</a>
         <a href="https://www.bbcrussian.com/news.lite">News</a>
-        <a href="/news.lite">News</a>
+        <a href="https://www.bbc.com/news.lite">News</a>
       `);
     });
 
diff --git a/src/server/Document/Renderers/litePageTransforms/transformAnchorTags.ts b/src/server/Document/Renderers/litePageTransforms/transformAnchorTags.ts
index c90f21f1a07..72ae7936a3d 100644
--- a/src/server/Document/Renderers/litePageTransforms/transformAnchorTags.ts
+++ b/src/server/Document/Renderers/litePageTransforms/transformAnchorTags.ts
@@ -15,26 +15,36 @@ const VALID_DOMAINS = [
 
 const RESERVED_ROUTE_EXTENSIONS = ['amp', 'app', 'lite'];
 
-const isValidHref = (href: string) => {
-  const url = new URL(href, 'http://localhost');
+const addLiteExtension = (href?: string) => {
+  if (!href) return null;
 
-  const extension = url?.pathname?.split('.')?.pop() || '';
-  const startsWithHash = href?.startsWith('#');
+  const url = new URL(href, getEnvConfig().SIMORGH_BASE_URL);
 
-  const hasReservedRouteExtension =
-    RESERVED_ROUTE_EXTENSIONS.includes(extension);
+  const extension = url.pathname?.split('.')?.pop() || '';
 
   const isValidDomain = VALID_DOMAINS.includes(url.hostname);
   const isWsService = SERVICES.includes(
-    url?.pathname?.split('/')?.[1] as Services,
+    url.pathname?.split('/')?.[1] as Services,
   );
 
-  return (
+  const startsWithHash = url.href?.startsWith('#');
+
+  const hasReservedRouteExtension =
+    RESERVED_ROUTE_EXTENSIONS.includes(extension);
+
+  const shouldAddLiteExtension =
     isValidDomain &&
     isWsService &&
     !hasReservedRouteExtension &&
-    !startsWithHash
-  );
+    !startsWithHash;
+
+  if (shouldAddLiteExtension) {
+    url.pathname += '.lite';
+
+    return url;
+  }
+
+  return null;
 };
 
 export default (html: string) => {
@@ -45,13 +55,12 @@ export default (html: string) => {
 
     anchorTags.forEach(tag => {
       const href = tag?.match(/href="([^"]*)"/)?.[1];
-      if (href && isValidHref(href)) {
-        const newUrl = new URL(href, getEnvConfig().SIMORGH_BASE_URL);
-        newUrl.pathname += '.lite';
+      const urlWithLite = addLiteExtension(href);
 
+      if (href && urlWithLite) {
         modifiedHtml = modifiedHtml.replace(
           tag,
-          tag.replace(href, newUrl.href),
+          tag.replace(href, urlWithLite.href),
         );
       }
     });