BUG: Fixes GH34411 _query_iterator now returns a list with an empty pd.DataFrame if chunksize is set and the resultset is empty

Author: JohanKahrstrom

pandas/io/sql.py

@@ -833,11 +833,17 @@ def _query_iterator(
         self, result, chunksize, columns, coerce_float=True, parse_dates=None
     ):
         """Return generator through chunked result set."""
+        has_read_data = False
         while True:
             data = result.fetchmany(chunksize)
             if not data:
+                if not has_read_data:
+                    yield DataFrame.from_records(
+                        [], columns=columns, coerce_float=coerce_float
+                    )
                 break
             else:
+                has_read_data = True
                 self.frame = DataFrame.from_records(
                     data, columns=columns, coerce_float=coerce_float
                 )
@@ -1227,11 +1233,21 @@ def _query_iterator(
         result, chunksize, columns, index_col=None, coerce_float=True, parse_dates=None
     ):
         """Return generator through chunked result set"""
+        has_read_data = False
         while True:
             data = result.fetchmany(chunksize)
             if not data:
+                if not has_read_data:
+                    yield _wrap_result(
+                        [],
+                        columns,
+                        index_col=index_col,
+                        coerce_float=coerce_float,
+                        parse_dates=parse_dates
+                    )
                 break
             else:
+                has_read_data = True
                 yield _wrap_result(
                     data,
                     columns,
@@ -1684,14 +1700,20 @@ def _query_iterator(
         cursor, chunksize, columns, index_col=None, coerce_float=True, parse_dates=None
     ):
         """Return generator through chunked result set"""
+        has_read_data = False
         while True:
             data = cursor.fetchmany(chunksize)
             if type(data) == tuple:
                 data = list(data)
             if not data:
                 cursor.close()
+                if not has_read_data:
+                    yield DataFrame.from_records(
+                        [], columns=columns, coerce_float=coerce_float
+                    )
                 break
             else:
+                has_read_data = True
                 yield _wrap_result(
                     data,
                     columns,

pandas/tests/io/test_sql.py

@@ -611,6 +611,14 @@ def test_read_sql_view(self):
         iris_frame = sql.read_sql_query("SELECT * FROM iris_view", self.conn)
         self._check_iris_loaded_frame(iris_frame)
 
+    def test_read_sql_with_chunksize_no_result(self):
+        with_batch = sql.read_sql_query("SELECT * FROM iris_view WHERE SepalLength < 0.0", self.conn, chunksize=5)
+        without_batch = sql.read_sql_query("SELECT * FROM iris_view WHERE SepalLength < 0.0", self.conn)
+        tm.assert_frame_equal(
+            pd.concat(with_batch),
+            without_batch
+        )
+
     def test_to_sql(self):
         sql.to_sql(self.test_frame1, "test_frame1", self.conn)
         assert sql.has_table("test_frame1", self.conn)