INTPYTHON-490 Fix segmentation fault when document doesn't match schema

Author: blink1073

bindings/python/pymongoarrow/lib.pyx

@@ -187,12 +187,12 @@ cdef class BuilderManager:
                 raise ValueError("Could not append raw value to", full_key.decode('utf8'))
 
             # Recurse into documents.
-            if value_t == BSON_TYPE_DOCUMENT:
+            if value_t == BSON_TYPE_DOCUMENT and builder.type_marker == BSON_TYPE_DOCUMENT:
                 bson_iter_recurse(doc_iter, &child_iter)
                 self.parse_document(&child_iter, full_key, BSON_TYPE_DOCUMENT)
 
             # Recurse into arrays.
-            if value_t == BSON_TYPE_ARRAY:
+            if value_t == BSON_TYPE_ARRAY and builder.type_marker == BSON_TYPE_ARRAY:
                 bson_iter_recurse(doc_iter, &child_iter)
                 self.parse_document(&child_iter, full_key, BSON_TYPE_ARRAY)
 

bindings/python/test/test_arrow.py

@@ -492,6 +492,12 @@ def test_schema_missing_field(self):
             out = func(self.coll, {} if func == find_arrow_all else [], schema=schema).drop(["_id"])
             self.assertEqual(out["list_field"].to_pylist(), expected)
 
+    def test_schema_incorrect_data_type(self):
+        self.coll.delete_many({})
+        self.coll.insert_one({"x": {"y": 1}})
+        out = find_arrow_all(self.coll, {}, schema=Schema({"x": str}))
+        assert out.to_pylist() == [{"x": None}]
+
     def test_auto_schema_nested(self):
         # Create table with random data of various types.
         _, data = self._create_nested_data()