From 9600b169ccdb7dda16d8d451f9227ce77274a009 Mon Sep 17 00:00:00 2001
From: dab246 <tdvu@linagora.com>
Date: Mon, 13 Oct 2025 10:40:56 +0700
Subject: [PATCH] TF-4054 Add comprehensive tests for `wrapPlainTextLinks`

---
 core/lib/utils/html/html_utils.dart  |  39 +++++--
 core/test/utils/html_utils_test.dart | 147 +++++++++++++++++++++++++++
 2 files changed, 179 insertions(+), 7 deletions(-)

diff --git a/core/lib/utils/html/html_utils.dart b/core/lib/utils/html/html_utils.dart
index 8ea5bf035..546b04fef 100644
--- a/core/lib/utils/html/html_utils.dart
+++ b/core/lib/utils/html/html_utils.dart
@@ -723,7 +723,12 @@ class HtmlUtils {
 
       if (container == null) return htmlString;
 
-      final urlRegex = RegExp(r'''(?:(?:https?:\/\/)|(?:www\.))[^\s<]+[^<.,:;\"')\]\s!?]''');
+      final urlRegex = RegExp(
+        r'''(?:(?:https?:\/\/)|(?:ftp:\/\/)|(?:mailto:)|(?:file:\/\/)|(?:www\.))(?!\.)(?!.*\.\.)([^\s<]+[^<.,:;\"\'\)\]\s!?])''',
+        caseSensitive: false,
+        multiLine: true,
+        dotAll: true,
+      );
 
       _processNode(container, urlRegex);
 
@@ -735,7 +740,16 @@ class HtmlUtils {
   }
 
   static final _skipTags = {
-    'a', 'img', 'video', 'audio', 'source', 'link', 'script', 'iframe'
+    'a',
+    'img',
+    'video',
+    'audio',
+    'source',
+    'link',
+    'script',
+    'iframe',
+    'code',
+    'pre',
   };
 
   static void _processNode(dom.Node node, RegExp urlRegex) {
@@ -763,11 +777,22 @@ class HtmlUtils {
             nodes.add(dom.Text(text.substring(lastIndex, start)));
           }
 
-          final href = url.startsWith('http') ? url : 'https://$url';
-          final link = dom.Element.tag('a')
-            ..attributes['href'] = href
-            ..text = url;
-          nodes.add(link);
+          // ignore unsafe URLs
+          if (url.toLowerCase().startsWith('javascript:') ||
+              url.toLowerCase().startsWith('data:')) {
+            nodes.add(dom.Text(url));
+          } else {
+            // Normalize href
+            final href = url.startsWith(RegExp(r'https?|ftp|mailto|file'))
+                ? url
+                : 'https://$url';
+
+            final link = dom.Element.tag('a')
+              ..attributes['href'] = href
+              ..text = url;
+
+            nodes.add(link);
+          }
 
           lastIndex = match.end;
         }
diff --git a/core/test/utils/html_utils_test.dart b/core/test/utils/html_utils_test.dart
index 180cee2cd..58d3353d7 100644
--- a/core/test/utils/html_utils_test.dart
+++ b/core/test/utils/html_utils_test.dart
@@ -698,5 +698,152 @@ void main() {
         reason: 'There must be no <a> tag inside href or src attributes',
       );
     });
+
+    test('Protocol variations: http/https/www', () {
+      const html = 'A https://example.com B http://foo.bar C www.site.com';
+      final result = HtmlUtils.wrapPlainTextLinks(html);
+
+      expect(result, contains('<a href="https://example.com">https://example.com</a>'));
+      expect(result, contains('<a href="http://foo.bar">http://foo.bar</a>'));
+      expect(result, contains('<a href="https://www.site.com">www.site.com</a>'));
+    });
+
+    test('Protocol variations: ftp, mailto, file', () {
+      const html =
+          'ftp://files.example.com mailto:me@example.com file:///Users/me/readme.txt';
+      final out = HtmlUtils.wrapPlainTextLinks(html);
+
+      expect(
+        out,
+        contains('<a href="ftp://files.example.com">ftp://files.example.com</a>'),
+      );
+
+      expect(
+        out,
+        contains('<a href="mailto:me@example.com">mailto:me@example.com</a>'),
+      );
+
+      expect(
+        out,
+        contains(
+          '<a href="file:///Users/me/readme.txt">file:///Users/me/readme.txt</a>',
+        ),
+      );
+    });
+
+    test('Fragment + query: keep the same href, do not double https://', () {
+      const html =
+          'View https://example.com/page?query=1#section and https://a.b/c#frag';
+      final out = HtmlUtils.wrapPlainTextLinks(html);
+
+      expect(
+        out,
+        contains(
+          '<a href="https://example.com/page?query=1#section">https://example.com/page?query=1#section</a>',
+        ),
+      );
+      expect(
+        out,
+        contains('<a href="https://a.b/c#frag">https://a.b/c#frag</a>'),
+      );
+    });
+
+    test('Do not wrap in <code> and <pre>', () {
+      const html = '<pre>https://example.com</pre> <code>www.site.com</code>';
+      final out = HtmlUtils.wrapPlainTextLinks(html);
+
+      expect(out, contains('<pre>https://example.com</pre>'));
+      expect(out, contains('<code>www.site.com</code>'));
+      expect(out, isNot(contains('<pre><a ')));
+      expect(out, isNot(contains('<code><a ')));
+    });
+
+    test('Do not wrap javascript: and data:', () {
+      const html = 'javascript:alert(1) and data:text/html;base64,AAAA';
+      final out = HtmlUtils.wrapPlainTextLinks(html);
+
+      expect(out, isNot(contains('<a ')));
+      expect(out, contains('javascript:alert(1)'));
+      expect(out, contains('data:text/html;base64,AAAA'));
+    });
+
+    test('Do not wrap URLs in attributes', () {
+      const html = '<img src="https://example.com/img.png">';
+      final out = HtmlUtils.wrapPlainTextLinks(html);
+
+      expect(out, equals('<img src="https://example.com/img.png">'));
+    });
+
+    test('IP addresses: http + www.*', () {
+      const html = 'http://192.168.1.1 www.127.0.0.1';
+      final out = HtmlUtils.wrapPlainTextLinks(html);
+
+      expect(out, contains('<a href="http://192.168.1.1">http://192.168.1.1</a>'));
+      expect(out, contains('<a href="https://www.127.0.0.1">www.127.0.0.1</a>'));
+    });
+
+    test('URLs in blockquote/list/table are wrapped', () {
+      const html = '''
+      <blockquote>https://example.com</blockquote>
+      <ul><li>www.site.com</li></ul>
+      <table><tr><td>https://abc.com</td></tr></table>
+      ''';
+      final out = HtmlUtils.wrapPlainTextLinks(html).replaceAll('\n', '');
+
+      expect(
+        out,
+        contains(
+          '<blockquote><a href="https://example.com">https://example.com</a></blockquote>',
+        ),
+      );
+      expect(
+        out,
+        contains('<li><a href="https://www.site.com">www.site.com</a></li>'),
+      );
+      expect(
+        out,
+        contains('<td><a href="https://abc.com">https://abc.com</a></td>'),
+      );
+    });
+
+    test('Malformed URLs: no wrap', () {
+      const html = 'Wrong: http//example and www..com';
+      final out = HtmlUtils.wrapPlainTextLinks(html);
+
+      expect(out, isNot(contains('<a href')));
+      expect(out, contains('http//example'));
+      expect(out, contains('www..com'));
+    });
+
+    test('Adjacent links without spaces: will now wrap into ONE link', () {
+      const html = 'www.example1.comwww.example2.com';
+      final out = HtmlUtils.wrapPlainTextLinks(html);
+
+      expect(
+        out,
+        contains(
+          '<a href="https://www.example1.comwww.example2.com">www.example1.comwww.example2.com</a>',
+        ),
+      );
+      expect(out.indexOf('<a '), greaterThanOrEqualTo(0));
+      expect(out.indexOf('<a ', out.indexOf('<a ') + 1), equals(-1));
+    });
+
+    test('Do not wrap when already in <a> (skip parent tag)', () {
+      const html =
+          '<a href="https://example.com">https://example.com</a> và www.site.com';
+      final out = HtmlUtils.wrapPlainTextLinks(html);
+
+      expect(
+        out,
+        contains(
+          '<a href="https://example.com">https://example.com</a>',
+        ),
+      );
+      expect(
+        out,
+        contains('<a href="https://www.site.com">www.site.com</a>'),
+      );
+    });
   });
 }
\ No newline at end of file