Normalize URL string ports per RFC 3986 section 3.2.3

CHANGES/1033.bugfix.rst

@@ -0,0 +1,4 @@
+Normalize default ports according to :rfc:`3986`.
+
+Specified ports are removed from the str representation of an URL if the port matches
+the schemes default port. -- by :user:`commonism`.

tests/test_url.py

@@ -219,6 +219,12 @@
     assert url.authority == "степан:пароль@слава.укр:8080"
 
 
+def test_authority_unknown_scheme():
+    v = "scheme://user:[email protected]:43/path/to?a=1&b=2"
+    url = URL(v)
+    assert str(url) == v
+
+
 def test_lowercase():
     url = URL("http://gitHUB.com")
     assert url.raw_host == "github.com"
@@ -1332,6 +1338,7 @@
 def test_is_default_port_for_absolute_url_with_default_port():
     url = URL("http://example.com:80")
     assert url.is_default_port()
+    assert str(url) == "http://example.com"
 
 
 def test_is_default_port_for_absolute_url_with_nondefault_port():

tests/test_url_update_netloc.py

@@ -191,14 +191,32 @@
     assert str(url.with_port(8888)) == "http://example.com:8888"
 
 
+def test_with_port_normalization():
+    url = URL("http://example.com")
+
+    assert str(url.with_scheme("https")) == "https://example.com"
+    assert str(url.with_scheme("https").with_port(443)) == "https://example.com"
+    assert str(url.with_port(443).with_scheme("https")) == "https://example.com"
+
+    u88 = url.with_port(88)
+    assert str(u88) == "http://example.com:88"
+    assert str(u88.with_port(80)) == "http://example.com"
+    assert str(u88.with_scheme("https")) == "https://example.com:88"
+
+    u80 = url.with_port(80)
+    assert str(u80) == "http://example.com"
+    assert str(u80.with_port(81)) == "http://example.com:81"
+    assert str(u80.with_scheme("https")) == "https://example.com:80"
+
+
 def test_with_port_with_no_port():
     url = URL("http://example.com")
     assert str(url.with_port(None)) == "http://example.com"
 
 
 def test_with_port_ipv6():
     url = URL("http://[::1]:8080/")
-    assert str(url.with_port(80)) == "http://[::1]:80/"
+    assert str(url.with_port(81)) == "http://[::1]:81/"
 
 
 def test_with_port_keeps_query_and_fragment():

yarl/_url.py

@@ -1,5 +1,6 @@
 import functools
 import math
+import socket
 import warnings
 from collections.abc import Mapping, Sequence
 from contextlib import suppress
@@ -291,6 +292,16 @@
         val = self._val
         if not val.path and self.is_absolute() and (val.query or val.fragment):
             val = val._replace(path="/")
+        if (port := self._get_port()) is None:
+            val = val._replace(
+                netloc=self._make_netloc(
+                    self.raw_user,
+                    self.raw_password,
+                    self.raw_host,
+                    port,
+                    encode_host=False,
+                )
+            )
         return urlunsplit(val)
 
     def __repr__(self):
@@ -382,10 +393,15 @@
         e.g. 'http://python.org' or 'http://python.org:80', False
         otherwise.
 
+        Return False for relative URLs
+
         """
-        if self.port is None:
-            return False
-        default = DEFAULT_PORTS.get(self.scheme)
+        if self.explicit_port is None:
+            """
+            a relative URL does not have an implicit port / default port
+            """
+            return self.port is not None
+        default = self._get_default_port()
         if default is None:
             return False
         return self.port == default
@@ -435,6 +451,21 @@
         """
         return self._val.netloc
 
+    def _get_default_port(self):
+        port = None
+        if self.scheme:
+            port = DEFAULT_PORTS.get(self.scheme)
+            if port is None:
+                with suppress(OSError):
+                    port = socket.getservbyname(self.scheme)
+        return port
+
+    def _get_port(self):
+        """Port or None if default port"""
+        if self._get_default_port() == self.port:
+            return None
+        return self.port
+
     @cached_property
     def authority(self):
         """Decoded authority part of URL.
@@ -727,9 +758,7 @@
                     f"Appending path {path!r} starting from slash is forbidden"
                 )
             path = path if encoded else self._PATH_QUOTER(path)
-            segments = [
-                segment for segment in reversed(path.split("/")) if segment != "."
-            ]
+            segments = list(reversed(path.split("/")))
             if not segments:
                 continue
             # remove trailing empty segment for all but the last path