Backport to 2.3.x: REGR: from_records not initializing subclasses properly (#60726) (#62436)

Co-authored-by: Patrick Hoefler <61934744+phofl@users.noreply.github.com>

Author: ianhi

doc/source/whatsnew/v2.3.3.rst

@@ -58,6 +58,12 @@ Bug fixes
 - The :meth:`DataFrame.iloc` now works correctly with ``copy_on_write`` option when assigning values after subsetting the columns of a homogeneous DataFrame (:issue:`60309`)
 
 
+Other Bug fixes
+~~~~~~~~~~~~~~~~
+
+- Fixed regression in :meth:`DataFrame.from_records` not initializing subclasses properly (:issue:`57008`)
+
+
 .. ---------------------------------------------------------------------------
 .. _whatsnew_233.contributors:
 

pandas/core/frame.py

@@ -2549,8 +2549,10 @@ def maybe_reorder(
 
         manager = _get_option("mode.data_manager", silent=True)
         mgr = arrays_to_mgr(arrays, columns, result_index, typ=manager)
-
-        return cls._from_mgr(mgr, axes=mgr.axes)
+        df = DataFrame._from_mgr(mgr, axes=mgr.axes)
+        if cls is not DataFrame:
+            return cls(df, copy=False)
+        return df
 
     def to_records(
         self, index: bool = True, column_dtypes=None, index_dtypes=None

pandas/tests/frame/test_subclass.py

@@ -784,6 +784,13 @@ def test_constructor_with_metadata():
     assert isinstance(subset, MySubclassWithMetadata)
 
 
+def test_constructor_with_metadata_from_records():
+    # GH#57008
+    df = MySubclassWithMetadata.from_records([{"a": 1, "b": 2}])
+    assert df.my_metadata is None
+    assert type(df) is MySubclassWithMetadata
+
+
 class SimpleDataFrameSubClass(DataFrame):
     """A subclass of DataFrame that does not define a constructor."""