Fix urls creation from richtext [STTNHUB-252] (superdesk#2481)

* Fix urls creation from richtext * add test * reformat code via black
MarkLark86 · Oct 23, 2023 · ce3862e · ce3862e
1 parent 527ecb0
commit ce3862e
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 4 deletions.
diff --git a/superdesk/text_utils.py b/superdesk/text_utils.py
@@ -212,7 +212,12 @@ def replacement(match_object):
             # so return the whole match.
             return match_object.group(0)
         else:
-            return '<a href="{0}" target="_blank">{1}</a>'.format(html.unescape(groups[1]), groups[1])
+            url = html.unescape(groups[1])
+            if url.startswith("www."):
+                url = "https://" + url
+            elif not url.startswith(("http://", "https://")):
+                url = "https://" + url
+            return '<a href="{0}" target="_blank">{1}</a>'.format(url, groups[1])
 
     value = EMAIL_REGEX.sub(r'<a href="mailto:\1">\1</a>', text)
     value = re.sub(URL_REGEX, replacement, value)

diff --git a/tests/text_utils_test.py b/tests/text_utils_test.py
@@ -141,7 +141,7 @@ def test_convert_plain_text_to_html(self):
         test_strings = [
             ["plain text", "<p>plain text</p>"],
             ["email me at [email protected]", '<p>email me at <a href="mailto:[email protected]">[email protected]</a></p>'],
-            ["find me on foobar.com", '<p>find me on <a href="foobar.com" target="_blank">foobar.com</a></p>'],
+            ["find me on foobar.com", '<p>find me on <a href="https://foobar.com" target="_blank">foobar.com</a></p>'],
             [
                 "https://www.foobar.com",
                 '<p><a href="https://www.foobar.com" target="_blank">https://www.foobar.com</a></p>',
@@ -152,7 +152,7 @@ def test_convert_plain_text_to_html(self):
                 "https://www.foobar.com/test?one=two&three=4</a></p>",
             ],
             ["https://foobar.com", '<p><a href="https://foobar.com" target="_blank">https://foobar.com</a></p>'],
-            ["www.foobar.com", '<p><a href="www.foobar.com" target="_blank">www.foobar.com</a></p>'],
+            ["www.foobar.com", '<p><a href="https://www.foobar.com" target="_blank">www.foobar.com</a></p>'],
             [
                 """Foo Bar
 
@@ -164,8 +164,12 @@ def test_convert_plain_text_to_html(self):
                 "<p>Foo Bar PTY LTD</p>"
                 "<p>Ph +61 23456789</p>"
                 '<p>Email: <a href="mailto:[email protected]">[email protected]</a> '
-                'Website: <a href="www.foobar.com" target="_blank">www.foobar.com</a></p>',
+                'Website: <a href="https://www.foobar.com" target="_blank">www.foobar.com</a></p>',
             ],
         ]
         for tests in test_strings:
             self.assertEqual(text_utils.plain_text_to_html(tests[0]), tests[1])
+
+        self.assertEqual(
+            text_utils.plain_text_to_html("foo.com"), '<p><a href="https://foo.com" target="_blank">foo.com</a></p>'
+        )