From d99f274320c006324ea0ac16955947b96d36f0f6 Mon Sep 17 00:00:00 2001
From: Thanos <111999343+Sachaa-Thanasius@users.noreply.github.com>
Date: Sat, 15 Jun 2024 12:44:26 -0400
Subject: [PATCH] Started annotating IRIReference and URIReference. -
 Substituted namedtuple inheritance with common base `typing.NamedTuple`
 subclass in misc.py, since these classes share almost the exact same
 interface. - Added a _typing_compat.py module to be able to import
 typing.Self, or a placeholder for it, in multiple other modules without
 bloating their code. - Added basic method annotations to the two reference
 classes. - Not annotations-related:  - Move the __hash__ implementation over
 to IRIReference from URIMixin to be congruent with URIReference.  - Made the
 __eq__ implementations more similar to avoid different behavior in cases of
 inheritance (rare as that might be).  - Added overloads to
 `normalizers.normalize_query` and `normalizers.normalize_fragment` to clearly
 indicate that None will get passed through. This behavior is relied upon by
 the library currently.

- Note: The runtime-related changes can be reverted and reattempted later if need be. Still passing all the tests currently.
---
 src/rfc3986/_mixin.py         |  2 --
 src/rfc3986/_typing_compat.py | 19 +++++++++++++++++
 src/rfc3986/builder.py        |  6 +++---
 src/rfc3986/iri.py            | 39 +++++++++++++++++++++++------------
 src/rfc3986/misc.py           | 13 +++++++++---
 src/rfc3986/normalizers.py    | 24 +++++++++++++++++++--
 src/rfc3986/uri.py            | 28 +++++++++++++++----------
 7 files changed, 97 insertions(+), 34 deletions(-)
 create mode 100644 src/rfc3986/_typing_compat.py

diff --git a/src/rfc3986/_mixin.py b/src/rfc3986/_mixin.py
index a6278c5..f55d13f 100644
--- a/src/rfc3986/_mixin.py
+++ b/src/rfc3986/_mixin.py
@@ -10,8 +10,6 @@
 class URIMixin:
     """Mixin with all shared methods for URIs and IRIs."""
 
-    __hash__ = tuple.__hash__
-
     def authority_info(self):
         """Return a dictionary with the ``userinfo``, ``host``, and ``port``.
 
diff --git a/src/rfc3986/_typing_compat.py b/src/rfc3986/_typing_compat.py
new file mode 100644
index 0000000..4822ecc
--- /dev/null
+++ b/src/rfc3986/_typing_compat.py
@@ -0,0 +1,19 @@
+import sys
+import typing as t
+
+__all__ = ("Self",)
+
+if sys.version_info >= (3, 11):
+    from typing import Self
+elif t.TYPE_CHECKING:
+    from typing_extensions import Self
+else:
+
+    class _PlaceholderMeta(type):
+        # This is meant to make it easier to debug the presence of placeholder
+        # classes.
+        def __repr__(self):
+            return f"placeholder for typing.{self.__name__}"
+
+    class Self(metaclass=_PlaceholderMeta):
+        """Placeholder for "typing.Self"."""
diff --git a/src/rfc3986/builder.py b/src/rfc3986/builder.py
index 9fc2ef8..42d4763 100644
--- a/src/rfc3986/builder.py
+++ b/src/rfc3986/builder.py
@@ -47,7 +47,7 @@ def __init__(
         scheme: t.Optional[str] = None,
         userinfo: t.Optional[str] = None,
         host: t.Optional[str] = None,
-        port: t.Optional[str] = None,
+        port: t.Optional[t.Union[int, str]] = None,
         path: t.Optional[str] = None,
         query: t.Optional[str] = None,
         fragment: t.Optional[str] = None,
@@ -60,7 +60,7 @@ def __init__(
             (optional)
         :param str host:
             (optional)
-        :param int port:
+        :param int | str port:
             (optional)
         :param str path:
             (optional)
@@ -72,7 +72,7 @@ def __init__(
         self.scheme = scheme
         self.userinfo = userinfo
         self.host = host
-        self.port = port
+        self.port = str(port) if port is not None else port
         self.path = path
         self.query = query
         self.fragment = fragment
diff --git a/src/rfc3986/iri.py b/src/rfc3986/iri.py
index 01f51ef..6578bc0 100644
--- a/src/rfc3986/iri.py
+++ b/src/rfc3986/iri.py
@@ -14,13 +14,13 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import typing as t
-from collections import namedtuple
 
 from . import compat
 from . import exceptions
 from . import misc
 from . import normalizers
 from . import uri
+from ._typing_compat import Self
 
 
 try:
@@ -29,9 +29,7 @@
     idna = None
 
 
-class IRIReference(
-    namedtuple("IRIReference", misc.URI_COMPONENTS), uri.URIMixin
-):
+class IRIReference(misc.URIReferenceBase, uri.URIMixin):
     """Immutable object representing a parsed IRI Reference.
 
     Can be encoded into an URIReference object via the procedure
@@ -42,10 +40,16 @@ class IRIReference(
         the future. Check for changes to the interface when upgrading.
     """
 
-    slots = ()
+    encoding: str
 
     def __new__(
-        cls, scheme, authority, path, query, fragment, encoding="utf-8"
+        cls,
+        scheme: t.Optional[str],
+        authority: t.Optional[str],
+        path: t.Optional[str],
+        query: t.Optional[str],
+        fragment: t.Optional[str],
+        encoding: str = "utf-8",
     ):
         """Create a new IRIReference."""
         ref = super().__new__(
@@ -59,14 +63,16 @@ def __new__(
         ref.encoding = encoding
         return ref
 
-    def __eq__(self, other):
+    __hash__ = tuple.__hash__
+
+    def __eq__(self, other: object):
         """Compare this reference to another."""
         other_ref = other
         if isinstance(other, tuple):
-            other_ref = self.__class__(*other)
+            other_ref = type(self)(*other)
         elif not isinstance(other, IRIReference):
             try:
-                other_ref = self.__class__.from_string(other)
+                other_ref = self.from_string(other)
             except TypeError:
                 raise TypeError(
                     "Unable to compare {}() to {}()".format(
@@ -77,7 +83,7 @@ def __eq__(self, other):
         # See http://tools.ietf.org/html/rfc3986#section-6.2
         return tuple(self) == tuple(other_ref)
 
-    def _match_subauthority(self):
+    def _match_subauthority(self) -> t.Optional[t.Match[str]]:
         return misc.ISUBAUTHORITY_MATCHER.match(self.authority)
 
     @classmethod
@@ -85,7 +91,7 @@ def from_string(
         cls,
         iri_string: t.Union[str, bytes, bytearray],
         encoding: str = "utf-8",
-    ):
+    ) -> Self:
         """Parse a IRI reference from the given unicode IRI string.
 
         :param str iri_string: Unicode IRI to be parsed into a reference.
@@ -104,7 +110,12 @@ def from_string(
             encoding,
         )
 
-    def encode(self, idna_encoder=None):  # noqa: C901
+    def encode(  # noqa: C901
+        self,
+        idna_encoder: t.Optional[  # pyright: ignore[reportRedeclaration]
+            t.Callable[[str], t.Union[str, bytes]]
+        ] = None,
+    ) -> "uri.URIReference":
         """Encode an IRIReference into a URIReference instance.
 
         If the ``idna`` module is installed or the ``rfc3986[idna]``
@@ -127,7 +138,9 @@ def encode(self, idna_encoder=None):  # noqa: C901
                         "and the IRI hostname requires encoding"
                     )
 
-                def idna_encoder(name):
+                def idna_encoder(name: str) -> t.Union[str, bytes]:
+                    assert idna  # Known to not be None at this point.
+
                     if any(ord(c) > 128 for c in name):
                         try:
                             return idna.encode(
diff --git a/src/rfc3986/misc.py b/src/rfc3986/misc.py
index 5e0d925..7489d8f 100644
--- a/src/rfc3986/misc.py
+++ b/src/rfc3986/misc.py
@@ -26,9 +26,16 @@
     # Break an import loop.
     from . import uri
 
-# These are enumerated for the named tuple used as a superclass of
-# URIReference
-URI_COMPONENTS = ["scheme", "authority", "path", "query", "fragment"]
+
+class URIReferenceBase(t.NamedTuple):
+    """The namedtuple used as a superclass of URIReference and IRIReference."""
+
+    scheme: t.Optional[str]
+    authority: t.Optional[str]
+    path: t.Optional[str]
+    query: t.Optional[str]
+    fragment: t.Optional[str]
+
 
 important_characters = {
     "generic_delimiters": abnf_regexp.GENERIC_DELIMITERS,
diff --git a/src/rfc3986/normalizers.py b/src/rfc3986/normalizers.py
index 902c23c..532bfaf 100644
--- a/src/rfc3986/normalizers.py
+++ b/src/rfc3986/normalizers.py
@@ -82,14 +82,34 @@ def normalize_path(path: str) -> str:
     return remove_dot_segments(path)
 
 
-def normalize_query(query: str) -> str:
+@t.overload
+def normalize_query(query: str) -> str:  # noqa: D103
+    ...
+
+
+@t.overload
+def normalize_query(query: None) -> None:  # noqa: D103
+    ...
+
+
+def normalize_query(query: t.Optional[str]) -> t.Optional[str]:
     """Normalize the query string."""
     if not query:
         return query
     return normalize_percent_characters(query)
 
 
-def normalize_fragment(fragment: str) -> str:
+@t.overload
+def normalize_fragment(fragment: str) -> str:  # noqa: D103
+    ...
+
+
+@t.overload
+def normalize_fragment(fragment: None) -> None:  # noqa: D103
+    ...
+
+
+def normalize_fragment(fragment: t.Optional[str]) -> t.Optional[str]:
     """Normalize the fragment string."""
     if not fragment:
         return fragment
diff --git a/src/rfc3986/uri.py b/src/rfc3986/uri.py
index f5448d3..6747447 100644
--- a/src/rfc3986/uri.py
+++ b/src/rfc3986/uri.py
@@ -14,15 +14,15 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import typing as t
-from collections import namedtuple
 
 from . import compat
 from . import misc
 from . import normalizers
 from ._mixin import URIMixin
+from ._typing_compat import Self
 
 
-class URIReference(namedtuple("URIReference", misc.URI_COMPONENTS), URIMixin):
+class URIReference(misc.URIReferenceBase, URIMixin):
     """Immutable object representing a parsed URI Reference.
 
     .. note::
@@ -80,10 +80,16 @@ class URIReference(namedtuple("URIReference", misc.URI_COMPONENTS), URIMixin):
         The port parsed from the authority.
     """
 
-    slots = ()
+    encoding: str
 
     def __new__(
-        cls, scheme, authority, path, query, fragment, encoding="utf-8"
+        cls,
+        scheme: t.Optional[str],
+        authority: t.Optional[str],
+        path: t.Optional[str],
+        query: t.Optional[str],
+        fragment: t.Optional[str],
+        encoding: str = "utf-8",
     ):
         """Create a new URIReference."""
         ref = super().__new__(
@@ -99,18 +105,18 @@ def __new__(
 
     __hash__ = tuple.__hash__
 
-    def __eq__(self, other):
+    def __eq__(self, other: object):
         """Compare this reference to another."""
         other_ref = other
         if isinstance(other, tuple):
-            other_ref = URIReference(*other)
+            other_ref = type(self)(*other)
         elif not isinstance(other, URIReference):
             try:
-                other_ref = URIReference.from_string(other)
+                other_ref = self.from_string(other)
             except TypeError:
                 raise TypeError(
-                    "Unable to compare URIReference() to {}()".format(
-                        type(other).__name__
+                    "Unable to compare {}() to {}()".format(
+                        type(self).__name__, type(other).__name__
                     )
                 )
 
@@ -118,7 +124,7 @@ def __eq__(self, other):
         naive_equality = tuple(self) == tuple(other_ref)
         return naive_equality or self.normalized_equality(other_ref)
 
-    def normalize(self):
+    def normalize(self) -> "URIReference":
         """Normalize this reference as described in Section 6.2.2.
 
         This is not an in-place normalization. Instead this creates a new
@@ -145,7 +151,7 @@ def from_string(
         cls,
         uri_string: t.Union[str, bytes, bytearray],
         encoding: str = "utf-8",
-    ):
+    ) -> Self:
         """Parse a URI reference from the given unicode URI string.
 
         :param str uri_string: Unicode URI to be parsed into a reference.