ARROW-176 Nested extension objects are not handled in auto schema

bindings/python/pymongoarrow/lib.pyx

@@ -26,7 +26,7 @@ from math import isnan
 # Python imports
 import bson
 import numpy as np
-from pyarrow import timestamp, struct, field
+from pyarrow import timestamp, struct, field, scalar, FixedSizeBinaryScalar, StructScalar
 from pyarrow.lib import (
     tobytes, StructType, int32, int64, float64, string, bool_, list_
 )
@@ -485,6 +485,9 @@ cdef class ObjectIdBuilder(_ArrayBuilderBase):
         cdef shared_ptr[CArray] out
         with nogil:
             self.builder.get().Finish(&out)
+        result = pyarrow_wrap_array(out)
+        for x in result:
+            print("CORRECT: ", x)
         return pyarrow_wrap_array(out).cast(ObjectIdType())
 
     cdef shared_ptr[CFixedSizeBinaryBuilder] unwrap(self):
@@ -781,7 +784,14 @@ cdef class DocumentBuilder(_ArrayBuilderBase):
         cdef shared_ptr[CArray] out
         with nogil:
             self.builder.get().Finish(&out)
-        return pyarrow_wrap_array(out)
+        wrapped = pyarrow_wrap_array(out)
+        python_out = []
+        for original in wrapped:
+            for fname, ftype in original.items():
+                if isinstance(ftype, FixedSizeBinaryScalar) and ftype.type.byte_width == 12: # ObjectIdType
+                    print("BEFORE: ", fname, ftype)
+                    print("AFTER: ", fname, ftype.cast(ObjectIdType()))
+        return wrapped
 
     cdef shared_ptr[CStructBuilder] unwrap(self):
         return self.builder

bindings/python/test/test_arrow.py

@@ -650,6 +650,15 @@ def test_nested_contradicting_unused_schema(self):
             out = func(self.coll, {} if func == find_arrow_all else [], schema=schema)
             self.assertEqual(out["obj"].to_pylist(), [{"a": 1}, {"a": 2}])
 
+    def test_nested_bson_objectId(self):
+        object_id = ObjectId()
+        data = {'_id': object_id, 'id1': object_id, 'obj': {'id2': object_id, 'id3': object_id}}
+
+        self.coll.drop()
+        self.coll.insert_one(data)
+        out = find_arrow_all(self.coll, {})
+        for schema_field in out.schema:
+            self.assertIsInstance(schema_field.type, ObjectIdType)
 
 class TestArrowExplicitApi(ArrowApiTestMixin, unittest.TestCase):
     def run_find(self, *args, **kwargs):