Fixed bug when fetching multiple consecutive null values into a data

frame.

Author: anthony-tuininga

doc/src/release_notes.rst

@@ -38,6 +38,7 @@ Thin Mode Changes
     maintenance not being sent when using connection pools with asyncio.
 #)  Fixed bug resulting in ``TypeError`` when using
     :attr:`DeqOptions.correlation` for buffered delivery mode.
+#)  Fixed bug when fetching multiple consecutive null values into a data frame.
 
 Thick Mode Changes
 ++++++++++++++++++

src/oracledb/interchange/nanoarrow_bridge.pyx

@@ -55,6 +55,9 @@ cdef extern from "nanoarrow/nanoarrow.c":
         ArrowBufferViewData data
         int64_t size_bytes
 
+    cdef struct ArrowBitmap:
+        ArrowBuffer buffer
+
     cdef struct ArrowArrayView:
         ArrowBufferView *buffer_views
 
@@ -88,6 +91,7 @@ cdef extern from "nanoarrow/nanoarrow.c":
     ArrowErrorCode ArrowArrayReserve(ArrowArray* array,
                                      int64_t additional_size_elements)
     ArrowErrorCode ArrowArrayStartAppending(ArrowArray* array)
+    ArrowBitmap* ArrowArrayValidityBitmap(ArrowArray* array)
     ArrowErrorCode ArrowArrayViewInitFromArray(ArrowArrayView* array_view,
                                                ArrowArray* array)
     int8_t ArrowBitGet(const uint8_t* bits, int64_t i)
@@ -335,8 +339,15 @@ cdef class OracleArrowArray:
             int64_t index
             uint8_t *ptr
             void* temp
+            ArrowBitmap *bitamp
         if array is None:
             array = self
+        bitmap = ArrowArrayValidityBitmap(array.arrow_array)
+        if bitmap != NULL and bitmap.buffer.data != NULL:
+            as_bool = ArrowBitGet(bitmap.buffer.data, index)
+            if not as_bool:
+                self.append_null()
+                return 0
         index = array.arrow_array.length - 1
         if array.arrow_type in (NANOARROW_TYPE_INT64, NANOARROW_TYPE_TIMESTAMP):
             data_buffer = ArrowArrayBuffer(array.arrow_array, 1)

tests/test_8000_dataframe.py

@@ -633,6 +633,45 @@ def test_8026(self):
         fetched_data = self.__get_data_from_df(fetched_df)
         self.assertEqual(fetched_data, data)
 
+    def test_8027(self):
+        "8027 - fetch data with multiple rows containing null values"
+        self.__check_interop()
+        ora_df = self.conn.fetch_df_all(
+            """
+            select to_date(null) as data from dual
+            union all
+            select to_date(null) as data from dual
+            union all
+            select to_date(null) as data from dual
+            union all
+            select to_date('2025-06-11', 'YYYY-MM-DD') as data from dual
+            union all
+            select to_date(null) as data from dual
+            union all
+            select to_date(null) as data from dual
+            union all
+            select to_date(null) as data from dual
+            union all
+            select to_date(null) as data from dual
+            """
+        )
+        data = [
+            (None,),
+            (None,),
+            (None,),
+            (datetime.datetime(2025, 6, 11),),
+            (None,),
+            (None,),
+            (None,),
+            (None,),
+        ]
+        fetched_tab = pyarrow.Table.from_arrays(
+            ora_df.column_arrays(), names=ora_df.column_names()
+        )
+        fetched_df = fetched_tab.to_pandas()
+        fetched_data = self.__get_data_from_df(fetched_df)
+        self.assertEqual(fetched_data, data)
+
 
 if __name__ == "__main__":
     test_env.run_test_cases()

tests/test_8100_dataframe_async.py

@@ -586,6 +586,45 @@ async def test_8122(self):
         fetched_data = self.__get_data_from_df(fetched_df)
         self.assertEqual(fetched_data, data)
 
+    async def test_8123(self):
+        "8123 - fetch data with multiple rows containing null values"
+        self.__check_interop()
+        ora_df = await self.conn.fetch_df_all(
+            """
+            select to_date(null) as data from dual
+            union all
+            select to_date(null) as data from dual
+            union all
+            select to_date(null) as data from dual
+            union all
+            select to_date('2025-06-11', 'YYYY-MM-DD') as data from dual
+            union all
+            select to_date(null) as data from dual
+            union all
+            select to_date(null) as data from dual
+            union all
+            select to_date(null) as data from dual
+            union all
+            select to_date(null) as data from dual
+            """
+        )
+        data = [
+            (None,),
+            (None,),
+            (None,),
+            (datetime.datetime(2025, 6, 11),),
+            (None,),
+            (None,),
+            (None,),
+            (None,),
+        ]
+        fetched_tab = pyarrow.Table.from_arrays(
+            ora_df.column_arrays(), names=ora_df.column_names()
+        )
+        fetched_df = fetched_tab.to_pandas()
+        fetched_data = self.__get_data_from_df(fetched_df)
+        self.assertEqual(fetched_data, data)
+
 
 if __name__ == "__main__":
     test_env.run_test_cases()