Merge pull request 'fix(previews): no genuine preview card for a site a deceptive link only pretends to be (#286)' (#404) from fix/preview-deceptive-links into lotus

src/app/components/message/MsgTypeRenderers.tsx

@@ -40,6 +40,7 @@ import { parseGeoUri, scaleYDimension } from '../../utils/common';
 import { formatAccuracy } from '../../utils/geoShare';
 import { Attachment, AttachmentBox, AttachmentContent, AttachmentHeader } from './attachment';
 import { FileHeader, FileDownloadButton } from './FileHeader';
+import { previewUrlsFor } from '../../utils/linkSafety';
 
 // ≈ 20 lines; in rem so a larger font setting doesn't show fewer lines (Gitea #343).
 const COLLAPSE_MAX_HEIGHT_REM = 20;
@@ -343,7 +344,13 @@ export function MText({
   if (typeof body !== 'string') return <BrokenContent />;
   const trimmedBody = trimReplyFromBody(body);
   const urlsMatch = renderUrlsPreview && trimmedBody.match(URL_REG);
-  const urls = urlsMatch ? [...new Set(urlsMatch)] : undefined;
+  // [Gitea #286] no genuine card for a site a deceptive link only pretends to be
+  const urls = urlsMatch
+    ? previewUrlsFor(
+        [...new Set(urlsMatch)],
+        typeof customBody === 'string' ? customBody : undefined,
+      )
+    : undefined;
 
   return (
     <>
@@ -401,7 +408,13 @@ export function MEmote({
   if (typeof body !== 'string') return <BrokenContent />;
   const trimmedBody = trimReplyFromBody(body);
   const urlsMatch = renderUrlsPreview && trimmedBody.match(URL_REG);
-  const urls = urlsMatch ? [...new Set(urlsMatch)] : undefined;
+  // [Gitea #286] no genuine card for a site a deceptive link only pretends to be
+  const urls = urlsMatch
+    ? previewUrlsFor(
+        [...new Set(urlsMatch)],
+        typeof customBody === 'string' ? customBody : undefined,
+      )
+    : undefined;
 
   return (
     <>
@@ -458,7 +471,13 @@ export function MNotice({
   if (typeof body !== 'string') return <BrokenContent />;
   const trimmedBody = trimReplyFromBody(body);
   const urlsMatch = renderUrlsPreview && trimmedBody.match(URL_REG);
-  const urls = urlsMatch ? [...new Set(urlsMatch)] : undefined;
+  // [Gitea #286] no genuine card for a site a deceptive link only pretends to be
+  const urls = urlsMatch
+    ? previewUrlsFor(
+        [...new Set(urlsMatch)],
+        typeof customBody === 'string' ? customBody : undefined,
+      )
+    : undefined;
 
   return (
     <>

src/app/utils/linkSafety.test.ts

@@ -1,6 +1,6 @@
 import { test } from 'node:test';
 import assert from 'node:assert/strict';
-import { analyzeLink, hostFromText, registrableDomain } from './linkSafety';
+import { analyzeLink, hostFromText, previewUrlsFor, registrableDomain } from './linkSafety';
 
 test('visible text that names another site is a mismatch', () => {
   const r = analyzeLink('https://matrix.lotusguild.org/login', 'https://evil.example/login');
@@ -73,3 +73,25 @@ test('shared hosting suffixes are separate sites (#335)', () => {
   );
   assert.equal(registrableDomain('news.bbc.co.uk'), 'bbc.co.uk');
 });
+
+test('previewUrlsFor: no preview for a site a deceptive link only claims (#286)', () => {
+  const urls = ['https://paypal.com'];
+  assert.deepEqual(
+    previewUrlsFor(urls, '<a href="https://evil.example/login">https://paypal.com</a>'),
+    [],
+  );
+  // An honest link keeps its preview.
+  assert.deepEqual(
+    previewUrlsFor(urls, '<a href="https://www.paypal.com/x">https://paypal.com</a>'),
+    urls,
+  );
+  assert.deepEqual(previewUrlsFor(urls, undefined), urls);
+  // Other links in the message keep theirs.
+  assert.deepEqual(
+    previewUrlsFor(
+      ['https://paypal.com', 'https://lotusguild.org'],
+      'see <a href="https://evil.example">paypal.com</a> and https://lotusguild.org',
+    ),
+    ['https://lotusguild.org'],
+  );
+});

src/app/utils/linkSafety.ts

@@ -221,3 +221,37 @@ export const analyzeLink = (text: string, href: string): LinkSafety | null => {
   const other = shown.find((h) => registrableDomain(h) !== realSite);
   return { mismatch: other !== undefined, shownHost: other ?? shown[0], realHost, punycode };
 };
+
+const ANCHOR_REG = /<a\s[^>]*href\s*=\s*(["'])(.*?)\1[^>]*>([\s\S]*?)<\/a>/gi;
+const decodeEntities = (s: string): string =>
+  s
+    .replace(/<[^>]+>/g, '')
+    .replace(/&quot;/g, '"')
+    .replace(/&#39;/g, "'")
+    .replace(/&lt;/g, '<')
+    .replace(/&gt;/g, '>')
+    .replace(/&amp;/g, '&');
+
+/**
+ * [Gitea #286] The URLs to build link previews for. Previews come from the
+ * plain `body`, but clicks follow `formatted_body`'s hrefs, and a sender
+ * controls both: text reading "https://paypal.com" linked to evil.example got
+ * a genuine PayPal card. Drop any body URL whose site is only *claimed* by a
+ * deceptive link's text.
+ */
+export const previewUrlsFor = (urls: string[], formattedBody: string | undefined): string[] => {
+  if (!formattedBody || urls.length === 0) return urls;
+  const claimed = new Set<string>();
+  Array.from(formattedBody.matchAll(ANCHOR_REG)).forEach((m) => {
+    const safety = analyzeLink(decodeEntities(m[3]), decodeEntities(m[2]));
+    if (safety?.mismatch && safety.shownHost) claimed.add(registrableDomain(safety.shownHost));
+  });
+  if (claimed.size === 0) return urls;
+  return urls.filter((u) => {
+    try {
+      return !claimed.has(registrableDomain(new URL(u).hostname));
+    } catch {
+      return true;
+    }
+  });
+};