Preserve non-default ports in Host headers (#6171)

Co-authored-by: cuteNBD <[email protected]>

lib/core/common.py

@@ -5138,12 +5138,12 @@ def getHostHeader(url):
     retVal = url
 
     if url:
-        retVal = _urllib.parse.urlparse(url).netloc
+        parsed = _urllib.parse.urlparse(url)
+        retVal = parsed.netloc
+        defaultPort = {"http": 80, "https": 443}.get(parsed.scheme.lower())
 
-        # Note: netloc keeps the IPv6 brackets (and any port), so only the default ports are
-        # stripped here - mirroring the hostname/IPv4 branch and preserving non-default ports
-        # (e.g. '[::1]:8080') as required by RFC 7230
-        if any(retVal.endswith(':%d' % _) for _ in (80, 443)):
+        # Keep non-default ports, including 80 on HTTPS and 443 on HTTP.
+        if defaultPort is not None and retVal.endswith(':%d' % defaultPort):
             retVal = retVal[:retVal.rfind(':')]
 
     if retVal and retVal.count(':') > 1 and not any(_ in retVal for _ in ('[', ']')):

tests/test_common_helpers.py

@@ -70,6 +70,21 @@ class TestGetHostHeader(unittest.TestCase):
     def test_without_port(self):
         self.assertEqual(getHostHeader("http://example.com/path"), "example.com")
 
+    def test_default_ports_depend_on_scheme(self):
+        cases = (
+            ("http://example.com:80/path", "example.com"),
+            ("https://example.com:443/path", "example.com"),
+            ("http://example.com:443/path", "example.com:443"),
+            ("https://example.com:80/path", "example.com:80"),
+            ("HTTP://example.com:443/path", "example.com:443"),
+            ("http://[::1]:80/path", "[::1]"),
+            ("https://[::1]:443/path", "[::1]"),
+            ("http://[::1]:443/path", "[::1]:443"),
+            ("https://[::1]:80/path", "[::1]:80"),
+        )
+        for url, expected in cases:
+            self.assertEqual(getHostHeader(url), expected, msg=url)
+
 
 if __name__ == "__main__":
     unittest.main(verbosity=2)