change protocol-relative urls to http

This commit is contained in:
selberhad 2022-10-07 22:19:58 -04:00
parent dfb623ce90
commit d6074c1386
2 changed files with 7 additions and 7 deletions

View file

@ -250,13 +250,13 @@ class TestTextToHTMLparser(TestCase):
def test_url_scheme_ftp(self):
self.assertEqual(
self.parser.convert_urls("ftp.example.com"),
'<a href="//ftp.example.com" target="_blank">ftp.example.com</a>',
'<a href="http://ftp.example.com" target="_blank">ftp.example.com</a>',
)
def test_url_scheme_www(self):
self.assertEqual(
self.parser.convert_urls("www.example.com"),
'<a href="//www.example.com" target="_blank">www.example.com</a>',
'<a href="http://www.example.com" target="_blank">www.example.com</a>',
)
def test_url_scheme_ftpproto(self):
@ -280,7 +280,7 @@ class TestTextToHTMLparser(TestCase):
def test_url_chars_slash(self):
self.assertEqual(
self.parser.convert_urls("www.example.com/homedir"),
'<a href="//www.example.com/homedir" target="_blank">www.example.com/homedir</a>',
'<a href="http://www.example.com/homedir" target="_blank">www.example.com/homedir</a>',
)
def test_url_chars_colon(self):
@ -316,13 +316,13 @@ class TestTextToHTMLparser(TestCase):
def test_url_edge_following_period_eol(self):
self.assertEqual(
self.parser.convert_urls("www.example.com."),
'<a href="//www.example.com" target="_blank">www.example.com</a>.',
'<a href="http://www.example.com" target="_blank">www.example.com</a>.',
)
def test_url_edge_following_period(self):
self.assertEqual(
self.parser.convert_urls("see www.example.com. "),
'see <a href="//www.example.com" target="_blank">www.example.com</a>. ',
'see <a href="http://www.example.com" target="_blank">www.example.com</a>. ',
)
def test_url_edge_brackets(self):

View file

@ -154,11 +154,11 @@ class TextToHTMLparser(object):
href = m.group(1)
label = href
# if there is no protocol (i.e. starts with www or ftp)
# prefix with // so the link isn't treated as relative
# prefix with http:// so the link isn't treated as relative
if not self.re_protocol.match(href):
if not self.re_valid_no_protocol.match(href):
return text
href = "//" + href
href = "http://" + href
rest = m.group(2)
# -> added target to output prevent the web browser from attempting to
# change pages (and losing our webclient session).