Give Pyright what it wants (alias attributes everywhere)

newsfragments/3114.bugfix.rst

@@ -0,0 +1 @@
+Ensure that Pyright recognizes our underscore prefixed attributes for attrs classes.

src/trio/_core/_local.py

@@ -38,8 +38,8 @@ class RunVar(Generic[T]):
 
     """
 
-    _name: str
-    _default: T | type[_NoValue] = _NoValue
+    _name: str = attrs.field(alias="name")
+    _default: T | type[_NoValue] = attrs.field(default=_NoValue, alias="default")
 
     def get(self, default: T | type[_NoValue] = _NoValue) -> T:
         """Gets the value of this :class:`RunVar` for the current run call."""

src/trio/_core/_run.py

@@ -544,9 +544,13 @@ class CancelScope:
     cancelled_caught: bool = attrs.field(default=False, init=False)
 
     # Constructor arguments:
-    _relative_deadline: float = attrs.field(default=inf, kw_only=True)
-    _deadline: float = attrs.field(default=inf, kw_only=True)
-    _shield: bool = attrs.field(default=False, kw_only=True)
+    _relative_deadline: float = attrs.field(
+        default=inf,
+        kw_only=True,
+        alias="relative_deadline",
+    )
+    _deadline: float = attrs.field(default=inf, kw_only=True, alias="deadline")
+    _shield: bool = attrs.field(default=False, kw_only=True, alias="shield")
 
     def __attrs_post_init__(self) -> None:
         if isnan(self._deadline):

src/trio/_tests/test_exports.py

@@ -9,6 +9,7 @@
 import json
 import socket as stdlib_socket
 import sys
+import tokenize
 import types
 from pathlib import Path, PurePath
 from types import ModuleType
@@ -572,3 +573,77 @@ def test_classes_are_final() -> None:
                 continue
 
             assert class_is_final(class_)
+
+
+def test_pyright_recognizes_init_attributes() -> None:
+    # Obviously, this isn't completely accurate.
+    # It should still be good enough. Hopefully.
+    # (attrs updates fields before we can access them)
+    files = []
+
+    parent = (Path(inspect.getfile(trio)) / "..").resolve()
+    for path in parent.glob("**/*.py"):
+        if "_tests" in str(path)[len(str(parent)) :]:
+            continue
+
+        with open(path, "rb") as f:
+            files.append(list(tokenize.tokenize(f.readline)))
+
+    for module in PUBLIC_MODULES:
+        for name, class_ in module.__dict__.items():
+            if not attrs.has(class_):
+                continue
+            if isinstance(class_, _util.NoPublicConstructor):
+                continue
+
+            file = None
+            start = None
+            for contents in files:
+                last_was_class = False
+                for i, token in enumerate(contents):
+                    if (
+                        token.type == tokenize.NAME
+                        and token.string == name
+                        and last_was_class
+                    ):
+                        assert file is None
+                        file = contents
+                        start = i - 1
+
+                    if token.type == tokenize.NAME and token.string == "class":
+                        last_was_class = True
+                    else:
+                        last_was_class = False
+
+            assert file is not None, f"{name}: {class_!r}"
+            assert start is not None
+
+            count = -1
+            # linters don't like my not using the index, go figure.
+            for end_offset, token in enumerate(file[start:]):  # noqa: B007
+                if token.type == tokenize.INDENT:
+                    count += 1
+                if token.type == tokenize.DEDENT and count:
+                    count -= 1
+                elif token.type == tokenize.DEDENT:
+                    break
+
+            assert token.type == tokenize.DEDENT
+            class_source = (
+                tokenize.untokenize(file[start : start + end_offset])
+                .replace("\\\n", "")
+                .strip()
+            )
+
+            attributes = list(attrs.fields(class_))
+            attributes = [attr for attr in attributes if attr.name.startswith("_")]
+            attributes = [attr for attr in attributes if attr.init]
+
+            attributes = [
+                # could this be improved by parsing AST? yes. this is simpler though.
+                attr
+                for attr in attributes
+                if f'alias="{attr.alias}"' not in class_source
+            ]
+
+            assert attributes == [], class_