From 08067e43c6fea4b090889cfea68f83ecbcdb6e75 Mon Sep 17 00:00:00 2001
From: "google-labs-jules[bot]"
 <161369871+google-labs-jules[bot]@users.noreply.github.com>
Date: Mon, 21 Apr 2025 22:33:12 +0000
Subject: [PATCH] Fix: Re-enable no-broken-links and ignore rel=nofollow

Re-enable the `no-broken-links` hint in the web-recommended
configuration. Update the hint logic to ignore `<a>` elements that
contain `rel="nofollow"` when checking for broken links (404, 410,
500, 503 status codes). This prevents reports for links explicitly
marked as not endorsed.

Add unit tests to verify that links with `rel="nofollow"` are
correctly skipped, while ensuring that regular broken links are
still reported as expected.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Fix #5754
---
 .../configuration-web-recommended/index.json  |  1 +
 packages/hint-no-broken-links/src/hint.ts     | 10 +++
 packages/hint-no-broken-links/tests/tests.ts  | 76 +++++++++++++++++++
 3 files changed, 87 insertions(+)

diff --git a/packages/configuration-web-recommended/index.json b/packages/configuration-web-recommended/index.json
index ab84c1dc472..8b0022e86b1 100644
--- a/packages/configuration-web-recommended/index.json
+++ b/packages/configuration-web-recommended/index.json
@@ -28,6 +28,7 @@
         "meta-charset-utf-8",
         "meta-viewport",
         "no-bom",
+        "no-broken-links",
         "no-disallowed-headers",
         "no-friendly-error-pages",
         "no-html-only-headers",
diff --git a/packages/hint-no-broken-links/src/hint.ts b/packages/hint-no-broken-links/src/hint.ts
index 6b7311789d2..3ffe5ba96ad 100644
--- a/packages/hint-no-broken-links/src/hint.ts
+++ b/packages/hint-no-broken-links/src/hint.ts
@@ -218,6 +218,16 @@ export default class NoBrokenLinksHint implements IHint {
             const requests: Promise<void>[] = [];
 
             for (const url of urls) {
+                // Skip anchor links with rel="nofollow".
+                if (element.nodeName === 'A') {
+                    const rel = element.getAttribute('rel');
+
+                    if (rel && rel.toLowerCase().includes('nofollow')) {
+                        debug(`Skipping URL ${url} due to rel="nofollow"`);
+                        continue;
+                    }
+                }
+
                 /*
                  * If the URL is not HTTP or HTTPS (e.g. `mailto:`),
                  * there is no need to validate.
diff --git a/packages/hint-no-broken-links/tests/tests.ts b/packages/hint-no-broken-links/tests/tests.ts
index b1a963a42a3..6dc1dff4530 100644
--- a/packages/hint-no-broken-links/tests/tests.ts
+++ b/packages/hint-no-broken-links/tests/tests.ts
@@ -91,6 +91,30 @@ const bodyWithInvalidDomainPreconnectLinkTag = `<div>
 <link rel="preconnect" href="https://invalid.domain/">
 </div>`;
 
+// Snippets for nofollow tests
+const bodyWithNoFollowBrokenLink = `<div>
+<a href='/404' rel='nofollow'>Broken NoFollow</a>
+</div>`;
+
+const bodyWithNoFollowValidLink = `<div>
+<a href='/valid' rel='nofollow'>Valid NoFollow</a>
+</div>`;
+
+const bodyWithNoFollowMultipleRel = `<div>
+<a href='/404' rel='noopener nofollow'>Broken NoFollow Multiple</a>
+</div>`;
+
+const bodyWithMixedLinks = `<div>
+<a href='/404-nofollow' rel='nofollow'>Broken NoFollow</a>
+<a href='/404-regular'>Broken Regular</a>
+</div>`;
+
+const bodyWithNoFollowAndRegularBroken = `<div>
+<a href='/404-anchor' rel='nofollow'>Broken Anchor NoFollow</a>
+<img src='/404.png'>
+</div>`;
+
+
 const tests: HintTest[] = [
     {
         name: `This test should pass as it has links with valid href value`,
@@ -288,6 +312,58 @@ const tests: HintTest[] = [
         }],
         serverConfig: generateHTMLPage('', bodyWithInvalidDomainPreconnectLinkTag)
     },
+    // New tests for rel="nofollow"
+    {
+        name: `Anchor with rel="nofollow" pointing to a 404 should not report an error`,
+        serverConfig: {
+            '/': { content: generateHTMLPage('', bodyWithNoFollowBrokenLink) },
+            '/404': { status: 404 }
+        }
+        // No reports expected
+    },
+    {
+        name: `Anchor with rel="nofollow" pointing to a valid resource should not report an error`,
+        serverConfig: {
+            '/': { content: generateHTMLPage('', bodyWithNoFollowValidLink) },
+            '/valid': { content: 'Valid page' }
+        }
+        // No reports expected
+    },
+    {
+        name: `Anchor with multiple rel values including "nofollow" pointing to a 404 should not report an error`,
+        serverConfig: {
+            '/': { content: generateHTMLPage('', bodyWithNoFollowMultipleRel) },
+            '/404': { status: 404 }
+        }
+        // No reports expected
+    },
+    {
+        name: `Mixed content: Regular broken link should report, nofollow broken link should not`,
+        reports: [{
+            message: `Broken link found (404 response).`,
+            severity: Severity.error,
+            location: { column: 50, line: 2 } // Location of the regular link
+        }],
+        serverConfig: {
+            '/': { content: generateHTMLPage('', bodyWithMixedLinks) },
+            '/404-nofollow': { status: 404 },
+            '/404-regular': { status: 404 }
+        }
+    },
+    {
+        name: `Mixed content: Broken image should report, nofollow broken anchor should not`,
+        reports: [{
+            message: `Broken link found (404 response).`,
+            severity: Severity.error,
+            location: { column: 8, line: 3 } // Location of the image tag
+        }],
+        serverConfig: {
+            '/': { content: generateHTMLPage('', bodyWithNoFollowAndRegularBroken) },
+            '/404-anchor': { status: 404 },
+            '/404.png': { status: 404 }
+        }
+    },
+    // End of new tests for rel="nofollow"
     {
         name: `This test should fail as it has a loop`,
         reports: [