Modernise type annotations using FA rules from ruff (#785)

* Modernise type annotations using `FA` rules from ruff

This enforces the use of `from __future__ import annotations` on all
files that use objects from `typing` that have clearer syntax available
in newer typing  annotation syntax.

* Use recursive types in `_parser`

This makes use of more accurate type checking enabled by newer type
checker versions.

---------

Co-authored-by: Pradyun Gedam <pradyunsg@users.noreply.github.com>

Author: pradyunsg

pyproject.toml

@@ -65,6 +65,7 @@ extend-select = [
     "B",
     "E",
     "F",
+    "FA",
     "I",
     "N",
     "UP",

src/packaging/_elffile.py

@@ -8,10 +8,12 @@
 ELF header: https://refspecs.linuxfoundation.org/elf/gabi4+/ch4.eheader.html
 """
 
+from __future__ import annotations
+
 import enum
 import os
 import struct
-from typing import IO, Optional, Tuple
+from typing import IO
 
 
 class ELFInvalid(ValueError):
@@ -87,11 +89,11 @@ def __init__(self, f: IO[bytes]) -> None:
         except struct.error as e:
             raise ELFInvalid("unable to parse machine and section information") from e
 
-    def _read(self, fmt: str) -> Tuple[int, ...]:
+    def _read(self, fmt: str) -> tuple[int, ...]:
         return struct.unpack(fmt, self._f.read(struct.calcsize(fmt)))
 
     @property
-    def interpreter(self) -> Optional[str]:
+    def interpreter(self) -> str | None:
         """
         The path recorded in the ``PT_INTERP`` section header.
         """

src/packaging/_manylinux.py

@@ -1,11 +1,13 @@
+from __future__ import annotations
+
 import collections
 import contextlib
 import functools
 import os
 import re
 import sys
 import warnings
-from typing import Dict, Generator, Iterator, NamedTuple, Optional, Sequence, Tuple
+from typing import Generator, Iterator, NamedTuple, Sequence
 
 from ._elffile import EIClass, EIData, ELFFile, EMachine
 
@@ -17,7 +19,7 @@
 # `os.PathLike` not a generic type until Python 3.9, so sticking with `str`
 # as the type for `path` until then.
 @contextlib.contextmanager
-def _parse_elf(path: str) -> Generator[Optional[ELFFile], None, None]:
+def _parse_elf(path: str) -> Generator[ELFFile | None, None, None]:
     try:
         with open(path, "rb") as f:
             yield ELFFile(f)
@@ -72,15 +74,15 @@ def _have_compatible_abi(executable: str, archs: Sequence[str]) -> bool:
 # For now, guess what the highest minor version might be, assume it will
 # be 50 for testing. Once this actually happens, update the dictionary
 # with the actual value.
-_LAST_GLIBC_MINOR: Dict[int, int] = collections.defaultdict(lambda: 50)
+_LAST_GLIBC_MINOR: dict[int, int] = collections.defaultdict(lambda: 50)
 
 
 class _GLibCVersion(NamedTuple):
     major: int
     minor: int
 
 
-def _glibc_version_string_confstr() -> Optional[str]:
+def _glibc_version_string_confstr() -> str | None:
     """
     Primary implementation of glibc_version_string using os.confstr.
     """
@@ -90,7 +92,7 @@ def _glibc_version_string_confstr() -> Optional[str]:
     # https://github.com/python/cpython/blob/fcf1d003bf4f0100c/Lib/platform.py#L175-L183
     try:
         # Should be a string like "glibc 2.17".
-        version_string: Optional[str] = os.confstr("CS_GNU_LIBC_VERSION")
+        version_string: str | None = os.confstr("CS_GNU_LIBC_VERSION")
         assert version_string is not None
         _, version = version_string.rsplit()
     except (AssertionError, AttributeError, OSError, ValueError):
@@ -99,7 +101,7 @@ def _glibc_version_string_confstr() -> Optional[str]:
     return version
 
 
-def _glibc_version_string_ctypes() -> Optional[str]:
+def _glibc_version_string_ctypes() -> str | None:
     """
     Fallback implementation of glibc_version_string using ctypes.
     """
@@ -143,12 +145,12 @@ def _glibc_version_string_ctypes() -> Optional[str]:
     return version_str
 
 
-def _glibc_version_string() -> Optional[str]:
+def _glibc_version_string() -> str | None:
     """Returns glibc version string, or None if not using glibc."""
     return _glibc_version_string_confstr() or _glibc_version_string_ctypes()
 
 
-def _parse_glibc_version(version_str: str) -> Tuple[int, int]:
+def _parse_glibc_version(version_str: str) -> tuple[int, int]:
     """Parse glibc version.
 
     We use a regexp instead of str.split because we want to discard any
@@ -168,7 +170,7 @@ def _parse_glibc_version(version_str: str) -> Tuple[int, int]:
 
 
 @functools.lru_cache
-def _get_glibc_version() -> Tuple[int, int]:
+def _get_glibc_version() -> tuple[int, int]:
     version_str = _glibc_version_string()
     if version_str is None:
         return (-1, -1)

src/packaging/_musllinux.py

@@ -4,11 +4,13 @@
 linked against musl, and what musl version is used.
 """
 
+from __future__ import annotations
+
 import functools
 import re
 import subprocess
 import sys
-from typing import Iterator, NamedTuple, Optional, Sequence
+from typing import Iterator, NamedTuple, Sequence
 
 from ._elffile import ELFFile
 
@@ -18,7 +20,7 @@ class _MuslVersion(NamedTuple):
     minor: int
 
 
-def _parse_musl_version(output: str) -> Optional[_MuslVersion]:
+def _parse_musl_version(output: str) -> _MuslVersion | None:
     lines = [n for n in (n.strip() for n in output.splitlines()) if n]
     if len(lines) < 2 or lines[0][:4] != "musl":
         return None
@@ -29,7 +31,7 @@ def _parse_musl_version(output: str) -> Optional[_MuslVersion]:
 
 
 @functools.lru_cache
-def _get_musl_version(executable: str) -> Optional[_MuslVersion]:
+def _get_musl_version(executable: str) -> _MuslVersion | None:
     """Detect currently-running musl runtime version.
 
     This is done by checking the specified executable's dynamic linking

src/packaging/_parser.py

@@ -4,8 +4,10 @@
 the implementation.
 """
 
+from __future__ import annotations
+
 import ast
-from typing import Any, List, NamedTuple, Optional, Tuple, Union
+from typing import NamedTuple, Sequence, Tuple, Union
 
 from ._tokenizer import DEFAULT_RULES, Tokenizer
 
@@ -41,20 +43,16 @@ def serialize(self) -> str:
 
 MarkerVar = Union[Variable, Value]
 MarkerItem = Tuple[MarkerVar, Op, MarkerVar]
-# MarkerAtom = Union[MarkerItem, List["MarkerAtom"]]
-# MarkerList = List[Union["MarkerList", MarkerAtom, str]]
-# mypy does not support recursive type definition
-# https://github.com/python/mypy/issues/731
-MarkerAtom = Any
-MarkerList = List[Any]
+MarkerAtom = Union[MarkerItem, Sequence["MarkerAtom"]]
+MarkerList = Sequence[Union["MarkerList", MarkerAtom, str]]
 
 
 class ParsedRequirement(NamedTuple):
     name: str
     url: str
-    extras: List[str]
+    extras: list[str]
     specifier: str
-    marker: Optional[MarkerList]
+    marker: MarkerList | None
 
 
 # --------------------------------------------------------------------------------------
@@ -87,7 +85,7 @@ def _parse_requirement(tokenizer: Tokenizer) -> ParsedRequirement:
 
 def _parse_requirement_details(
     tokenizer: Tokenizer,
-) -> Tuple[str, str, Optional[MarkerList]]:
+) -> tuple[str, str, MarkerList | None]:
     """
     requirement_details = AT URL (WS requirement_marker?)?
                         | specifier WS? (requirement_marker)?
@@ -156,7 +154,7 @@ def _parse_requirement_marker(
     return marker
 
 
-def _parse_extras(tokenizer: Tokenizer) -> List[str]:
+def _parse_extras(tokenizer: Tokenizer) -> list[str]:
     """
     extras = (LEFT_BRACKET wsp* extras_list? wsp* RIGHT_BRACKET)?
     """
@@ -175,11 +173,11 @@ def _parse_extras(tokenizer: Tokenizer) -> List[str]:
     return extras
 
 
-def _parse_extras_list(tokenizer: Tokenizer) -> List[str]:
+def _parse_extras_list(tokenizer: Tokenizer) -> list[str]:
     """
     extras_list = identifier (wsp* ',' wsp* identifier)*
     """
-    extras: List[str] = []
+    extras: list[str] = []
 
     if not tokenizer.check("IDENTIFIER"):
         return extras

src/packaging/_tokenizer.py

@@ -1,7 +1,9 @@
+from __future__ import annotations
+
 import contextlib
 import re
 from dataclasses import dataclass
-from typing import Dict, Iterator, NoReturn, Optional, Tuple, Union
+from typing import Iterator, NoReturn
 
 from .specifiers import Specifier
 
@@ -21,7 +23,7 @@ def __init__(
         message: str,
         *,
         source: str,
-        span: Tuple[int, int],
+        span: tuple[int, int],
     ) -> None:
         self.span = span
         self.message = message
@@ -34,7 +36,7 @@ def __str__(self) -> str:
         return "\n    ".join([self.message, self.source, marker])
 
 
-DEFAULT_RULES: "Dict[str, Union[str, re.Pattern[str]]]" = {
+DEFAULT_RULES: dict[str, str | re.Pattern[str]] = {
     "LEFT_PARENTHESIS": r"\(",
     "RIGHT_PARENTHESIS": r"\)",
     "LEFT_BRACKET": r"\[",
@@ -96,13 +98,13 @@ def __init__(
         self,
         source: str,
         *,
-        rules: "Dict[str, Union[str, re.Pattern[str]]]",
+        rules: dict[str, str | re.Pattern[str]],
     ) -> None:
         self.source = source
-        self.rules: Dict[str, re.Pattern[str]] = {
+        self.rules: dict[str, re.Pattern[str]] = {
             name: re.compile(pattern) for name, pattern in rules.items()
         }
-        self.next_token: Optional[Token] = None
+        self.next_token: Token | None = None
         self.position = 0
 
     def consume(self, name: str) -> None:
@@ -154,8 +156,8 @@ def raise_syntax_error(
         self,
         message: str,
         *,
-        span_start: Optional[int] = None,
-        span_end: Optional[int] = None,
+        span_start: int | None = None,
+        span_end: int | None = None,
     ) -> NoReturn:
         """Raise ParserSyntaxError at the given position."""
         span = (

src/packaging/markers.py

@@ -2,22 +2,16 @@
 # 2.0, and the BSD License. See the LICENSE file in the root of this repository
 # for complete details.
 
+from __future__ import annotations
+
 import operator
 import os
 import platform
 import sys
-from typing import Any, Callable, Dict, List, Optional, Tuple, Union
-
-from ._parser import (
-    MarkerAtom,
-    MarkerList,
-    Op,
-    Value,
-    Variable,
-)
-from ._parser import (
-    parse_marker as _parse_marker,
-)
+from typing import Any, Callable
+
+from ._parser import MarkerAtom, MarkerList, Op, Value, Variable
+from ._parser import parse_marker as _parse_marker
 from ._tokenizer import ParserSyntaxError
 from .specifiers import InvalidSpecifier, Specifier
 from .utils import canonicalize_name
@@ -69,7 +63,7 @@ def _normalize_extra_values(results: Any) -> Any:
 
 
 def _format_marker(
-    marker: Union[List[str], MarkerAtom, str], first: Optional[bool] = True
+    marker: list[str] | MarkerAtom | str, first: bool | None = True
 ) -> str:
     assert isinstance(marker, (list, tuple, str))
 
@@ -96,7 +90,7 @@ def _format_marker(
         return marker
 
 
-_operators: Dict[str, Operator] = {
+_operators: dict[str, Operator] = {
     "in": lambda lhs, rhs: lhs in rhs,
     "not in": lambda lhs, rhs: lhs not in rhs,
     "<": operator.lt,
@@ -116,14 +110,14 @@ def _eval_op(lhs: str, op: Op, rhs: str) -> bool:
     else:
         return spec.contains(lhs, prereleases=True)
 
-    oper: Optional[Operator] = _operators.get(op.serialize())
+    oper: Operator | None = _operators.get(op.serialize())
     if oper is None:
         raise UndefinedComparison(f"Undefined {op!r} on {lhs!r} and {rhs!r}.")
 
     return oper(lhs, rhs)
 
 
-def _normalize(*values: str, key: str) -> Tuple[str, ...]:
+def _normalize(*values: str, key: str) -> tuple[str, ...]:
     # PEP 685 – Comparison of extra names for optional distribution dependencies
     # https://peps.python.org/pep-0685/
     # > When comparing extra names, tools MUST normalize the names being
@@ -135,8 +129,8 @@ def _normalize(*values: str, key: str) -> Tuple[str, ...]:
     return values
 
 
-def _evaluate_markers(markers: MarkerList, environment: Dict[str, str]) -> bool:
-    groups: List[List[bool]] = [[]]
+def _evaluate_markers(markers: MarkerList, environment: dict[str, str]) -> bool:
+    groups: list[list[bool]] = [[]]
 
     for marker in markers:
         assert isinstance(marker, (list, tuple, str))
@@ -165,15 +159,15 @@ def _evaluate_markers(markers: MarkerList, environment: Dict[str, str]) -> bool:
     return any(all(item) for item in groups)
 
 
-def format_full_version(info: "sys._version_info") -> str:
+def format_full_version(info: sys._version_info) -> str:
     version = "{0.major}.{0.minor}.{0.micro}".format(info)
     kind = info.releaselevel
     if kind != "final":
         version += kind[0] + str(info.serial)
     return version
 
 
-def default_environment() -> Dict[str, str]:
+def default_environment() -> dict[str, str]:
     iver = format_full_version(sys.implementation.version)
     implementation_name = sys.implementation.name
     return {
@@ -232,7 +226,7 @@ def __eq__(self, other: Any) -> bool:
 
         return str(self) == str(other)
 
-    def evaluate(self, environment: Optional[Dict[str, str]] = None) -> bool:
+    def evaluate(self, environment: dict[str, str] | None = None) -> bool:
         """Evaluate a marker.
 
         Return the boolean from evaluating the given marker against the