[ty] Override __file__ to str when applicable on imported modules (#22333)

crates/ty_python_semantic/resources/mdtest/import/dunder_file_attribute.md

@@ -0,0 +1,84 @@
+# The `__file__` attribute on imported modules
+
+## Module successfully resolved
+
+```py
+from b import __file__ as module_path
+from stub import __file__ as stub_path
+from override import __file__ as overidden_path
+
+reveal_type(__file__)  # revealed: str
+reveal_type(module_path)  # revealed: str
+reveal_type(stub_path)  # revealed: str
+reveal_type(overidden_path)  # revealed: None
+
+# NOTE: This import fails at runtime as this is a C Extension
+# with no `__file__` global. It's hard for us to determine this
+# right now, however (all we know is it comes from a stub file),
+# and if it did exist then it would be of type `str` since `sys`
+# is not a namespace packages. This behaviour also matches other
+# type checkers.
+from sys import __file__ as no_path
+
+reveal_type(no_path)  # revealed: str
+```
+
+`b.py`:
+
+```py
+
+```
+
+`override.py`:
+
+```py
+__file__ = None
+```
+
+`stub.pyi`:
+
+```pyi
+```
+
+## Module resolution failed
+
+```py
+from bar import __file__ as module_path  # error: "Cannot resolve imported module `bar`"
+
+reveal_type(module_path)  # revealed: Unknown
+```
+
+## Non-namespace packages have `__file__` available
+
+`a/__init__.py`:
+
+```py
+```
+
+`a/b.py`:
+
+```py
+```
+
+```py
+import a
+
+reveal_type(a.__file__)  # revealed: str
+```
+
+## `__file__` is set to `None` for namespace packages
+
+`namespace/c.py`:
+
+```py
+```
+
+```py
+import namespace
+
+# TODO: `__file__` does exist on namespace packages but is set to `None`;
+# this is a false positive
+#
+# error: [unresolved-attribute] "Module `namespace` has no member `__file__`"
+reveal_type(namespace.__file__)  # revealed: Unknown
+```

crates/ty_python_semantic/resources/mdtest/scopes/moduletype_attrs.md

@@ -99,9 +99,14 @@ import typing
 reveal_type(typing.__name__)  # revealed: str
 reveal_type(typing.__init__)  # revealed: bound method ModuleType.__init__(name: str, doc: str | None = ...) -> None
 
-# For a stub module, we don't know that `__file__` is a string (at runtime it may be entirely
-# unset, but we follow typeshed here):
-reveal_type(typing.__file__)  # revealed: str | None
+# Note that since the source for the `typing` module is a stub file,
+# we can't know for sure that it's not a C extension at runtime,
+# and C extensions don't necessarily have a `__file__` global attribute
+# at all (in which case this attribute access would fail). However, we
+# *do* know that `typing` is not a namespace package, so if `__file__`
+# does exist, it will be of type `str` (`__file__` is only `None` for
+# namespace packages).
+reveal_type(typing.__file__)  # revealed: str
 
 # These come from `builtins.object`, not `types.ModuleType`:
 reveal_type(typing.__eq__)  # revealed: bound method ModuleType.__eq__(value: object, /) -> bool

crates/ty_python_semantic/src/place.rs

@@ -1098,6 +1098,18 @@ fn symbol_impl<'db>(
     considered_definitions: ConsideredDefinitions,
 ) -> PlaceAndQualifiers<'db> {
     let _span = tracing::trace_span!("symbol", ?name).entered();
+    let place = place_table(db, scope)
+        .symbol_id(name)
+        .map(|symbol| {
+            place_by_id(
+                db,
+                scope,
+                symbol.into(),
+                requires_explicit_reexport,
+                considered_definitions,
+            )
+        })
+        .unwrap_or_default();
 
     if name == "platform"
         && file_to_module(db, scope.file(db))
@@ -1111,20 +1123,43 @@ fn symbol_impl<'db>(
                 // Fall through to the looked up type
             }
         }
+    } else if name == "__file__"
+        && let Some(module) = file_to_module(db, scope.file(db))
+    {
+        // We special-case `__file__` here because we know that for a successfully imported
+        // non-namespace-package Python module, that hasn't been explicitly overridden it
+        // is always a string, even though typeshed says `str | None`. For a namespace package,
+        // meanwhile, it will always be `None`.
+        //
+        // Note that C-extension modules (stdlib examples include `sys`, `itertools`, etc.)
+        //  may not have a `__file__` attribute at runtime at all, but that doesn't really
+        // affect the *type* of the attribute, just the *boundness*. There's no way for us
+        // to know right now whether a stub represents a C extension or not, so for now we
+        // do not attempt to detect this; we just infer `str` still. This matches the
+        // behaviour of other major type checkers.
+        let default_type = if module.file(db).is_some() {
+            KnownClass::Str.to_instance(db)
+        } else {
+            Type::none(db)
+        };
+
+        match place.place {
+            Place::Defined(defined_place) => match defined_place.definedness {
+                Definedness::AlwaysDefined => return place,
+                Definedness::PossiblyUndefined => {
+                    let new_type = UnionType::from_elements(db, [defined_place.ty, default_type]);
+                    let def_place = DefinedPlace::new(new_type)
+                        .with_definedness(Definedness::AlwaysDefined)
+                        .with_widening(defined_place.widening)
+                        .with_origin(defined_place.origin);
+                    return Place::Defined(def_place).into();
+                }
+            },
+            Place::Undefined => return Place::bound(default_type).into(),
+        }
     }
 
-    place_table(db, scope)
-        .symbol_id(name)
-        .map(|symbol| {
-            place_by_id(
-                db,
-                scope,
-                symbol.into(),
-                requires_explicit_reexport,
-                considered_definitions,
-            )
-        })
-        .unwrap_or_default()
+    place
 }
 
 fn place_impl<'db>(