Fix PyFunction doc behavior (#5827)

youknowone · web-flow · commit 52301ddbe549 · 2025-06-23T22:37:51.000+09:00
diff --git a/Lib/test/test_funcattrs.py b/Lib/test/test_funcattrs.py
@@ -357,8 +357,6 @@ def test_delete___dict__(self):
         else:
             self.fail("deleting function dictionary should raise TypeError")
 
-    # TODO: RUSTPYTHON
-    @unittest.expectedFailure
     def test_unassigned_dict(self):
         self.assertEqual(self.b.__dict__, {})
 
@@ -379,8 +377,6 @@ def test_set_docstring_attr(self):
         self.assertEqual(self.fi.a.__doc__, docstr)
         self.cannot_set_attr(self.fi.a, "__doc__", docstr, AttributeError)
 
-    # TODO: RUSTPYTHON
-    @unittest.expectedFailure
     def test_delete_docstring(self):
         self.b.__doc__ = "The docstring"
         del self.b.__doc__
diff --git a/extra_tests/snippets/syntax_function2.py b/extra_tests/snippets/syntax_function2.py
@@ -52,6 +52,8 @@ def f4():
 
 assert f4.__doc__ == "test4"
 
+assert type(lambda: None).__doc__.startswith("Create a function object."), type(f4).__doc__
+
 
 def revdocstr(f):
     d = f.__doc__
diff --git a/vm/src/builtins/descriptor.rs b/vm/src/builtins/descriptor.rs
@@ -345,15 +345,28 @@ impl GetDescriptor for PyMemberDescriptor {
     fn descr_get(
         zelf: PyObjectRef,
         obj: Option<PyObjectRef>,
-        _cls: Option<PyObjectRef>,
+        cls: Option<PyObjectRef>,
         vm: &VirtualMachine,
     ) -> PyResult {
+        let descr = Self::_as_pyref(&zelf, vm)?;
         match obj {
-            Some(x) => {
-                let zelf = Self::_as_pyref(&zelf, vm)?;
-                zelf.member.get(x, vm)
+            Some(x) => descr.member.get(x, vm),
+            None => {
+                // When accessed from class (not instance), for __doc__ member descriptor,
+                // return the class's docstring if available
+                // When accessed from class (not instance), check if the class has
+                // an attribute with the same name as this member descriptor
+                if let Some(cls) = cls {
+                    if let Ok(cls_type) = cls.downcast::<PyType>() {
+                        if let Some(interned) = vm.ctx.interned_str(descr.member.name.as_str()) {
+                            if let Some(attr) = cls_type.attributes.read().get(&interned) {
+                                return Ok(attr.clone());
+                            }
+                        }
+                    }
+                }
+                Ok(zelf)
             }
-            None => Ok(zelf),
         }
     }
 }
diff --git a/vm/src/builtins/function.rs b/vm/src/builtins/function.rs
@@ -417,10 +417,15 @@ impl PyFunction {
     }
 
     #[pymember(magic)]
-    fn doc(_vm: &VirtualMachine, zelf: PyObjectRef) -> PyResult {
-        let zelf: PyRef<PyFunction> = zelf.downcast().unwrap_or_else(|_| unreachable!());
-        let doc = zelf.doc.lock();
-        Ok(doc.clone())
+    fn doc(vm: &VirtualMachine, obj: PyObjectRef) -> PyResult {
+        // When accessed from instance, obj is the PyFunction instance
+        if let Ok(func) = obj.downcast::<PyFunction>() {
+            let doc = func.doc.lock();
+            Ok(doc.clone())
+        } else {
+            // When accessed from class, return None as there's no instance
+            Ok(vm.ctx.none())
+        }
     }
 
     #[pymember(magic, setter)]
diff --git a/vm/src/builtins/type.rs b/vm/src/builtins/type.rs
@@ -1061,6 +1061,22 @@ pub(crate) fn get_text_signature_from_internal_doc<'a>(
     find_signature(name, internal_doc).and_then(get_signature)
 }
 
+// _PyType_GetDocFromInternalDoc in CPython
+fn get_doc_from_internal_doc<'a>(name: &str, internal_doc: &'a str) -> &'a str {
+    // Similar to CPython's _PyType_DocWithoutSignature
+    // If the doc starts with the type name and a '(', it's a signature
+    if let Some(doc_without_sig) = find_signature(name, internal_doc) {
+        // Find where the signature ends
+        if let Some(sig_end_pos) = doc_without_sig.find(SIGNATURE_END_MARKER) {
+            let after_sig = &doc_without_sig[sig_end_pos + SIGNATURE_END_MARKER.len()..];
+            // Return the documentation after the signature, or empty string if none
+            return after_sig;
+        }
+    }
+    // If no signature found, return the whole doc
+    internal_doc
+}
+
 impl GetAttr for PyType {
     fn getattro(zelf: &Py<Self>, name_str: &Py<PyStr>, vm: &VirtualMachine) -> PyResult {
         #[cold]
@@ -1122,6 +1138,55 @@ impl Py<PyType> {
         PyTuple::new_unchecked(elements.into_boxed_slice())
     }
 
+    #[pygetset(magic)]
+    fn doc(&self, vm: &VirtualMachine) -> PyResult {
+        // Similar to CPython's type_get_doc
+        // For non-heap types (static types), check if there's an internal doc
+        if !self.slots.flags.has_feature(PyTypeFlags::HEAPTYPE) {
+            if let Some(internal_doc) = self.slots.doc {
+                // Process internal doc, removing signature if present
+                let doc_str = get_doc_from_internal_doc(&self.name(), internal_doc);
+                return Ok(vm.ctx.new_str(doc_str).into());
+            }
+        }
+
+        // Check if there's a __doc__ in the type's dict
+        if let Some(doc_attr) = self.get_attr(vm.ctx.intern_str("__doc__")) {
+            // If it's a descriptor, call its __get__ method
+            let descr_get = doc_attr
+                .class()
+                .mro_find_map(|cls| cls.slots.descr_get.load());
+            if let Some(descr_get) = descr_get {
+                descr_get(doc_attr, None, Some(self.to_owned().into()), vm)
+            } else {
+                Ok(doc_attr)
+            }
+        } else {
+            Ok(vm.ctx.none())
+        }
+    }
+
+    #[pygetset(magic, setter)]
+    fn set_doc(&self, value: PySetterValue, vm: &VirtualMachine) -> PyResult<()> {
+        // Similar to CPython's type_set_doc
+        let value = value.ok_or_else(|| {
+            vm.new_type_error(format!(
+                "cannot delete '__doc__' attribute of type '{}'",
+                self.name()
+            ))
+        })?;
+
+        // Check if we can set this special type attribute
+        self.check_set_special_type_attr(&value, identifier!(vm, __doc__), vm)?;
+
+        // Set the __doc__ in the type's dict
+        self.attributes
+            .write()
+            .insert(identifier!(vm, __doc__), value);
+
+        Ok(())
+    }
+
     #[pymethod(magic)]
     fn dir(&self) -> PyList {
         let attributes: Vec<PyObjectRef> = self
diff --git a/vm/src/class.rs b/vm/src/class.rs
@@ -96,7 +96,12 @@ pub trait PyClassImpl: PyClassDef {
         }
         Self::impl_extend_class(ctx, class);
         if let Some(doc) = Self::DOC {
-            class.set_attr(identifier!(ctx, __doc__), ctx.new_str(doc).into());
+            // Only set __doc__ if it doesn't already exist (e.g., as a member descriptor)
+            // This matches CPython's behavior in type_dict_set_doc
+            let doc_attr_name = identifier!(ctx, __doc__);
+            if class.attributes.read().get(doc_attr_name).is_none() {
+                class.set_attr(doc_attr_name, ctx.new_str(doc).into());
+            }
         }
         if let Some(module_name) = Self::MODULE_NAME {
             class.set_attr(

Original file line number	Diff line number	Diff line change
`@@ -96,7 +96,12 @@ pub trait PyClassImpl: PyClassDef {`
`96`	`96`	`}`
`97`	`97`	`Self::impl_extend_class(ctx, class);`
`98`	`98`	`if let Some(doc) = Self::DOC {`
`99`		`- class.set_attr(identifier!(ctx, __doc__), ctx.new_str(doc).into());`
	`99`	`+ // Only set __doc__ if it doesn't already exist (e.g., as a member descriptor)`
	`100`	`+ // This matches CPython's behavior in type_dict_set_doc`
	`101`	`+ let doc_attr_name = identifier!(ctx, __doc__);`
	`102`	`+ if class.attributes.read().get(doc_attr_name).is_none() {`
	`103`	`+ class.set_attr(doc_attr_name, ctx.new_str(doc).into());`
	`104`	`+ }`
`100`	`105`	`}`
`101`	`106`	`if let Some(module_name) = Self::MODULE_NAME {`
`102`	`107`	`class.set_attr(`