Fix read_sql empty result with chunksize bug GH34411 (#34429)

Author: JohanKahrstrom

doc/source/whatsnew/v1.3.0.rst

@@ -281,6 +281,7 @@ I/O
 - :func:`read_excel` now respects :func:`set_option` (:issue:`34252`)
 - Bug in :func:`read_csv` not switching ``true_values`` and ``false_values`` for nullable ``boolean`` dtype (:issue:`34655`)
 - Bug in :func:`read_json` when ``orient="split"`` does not maintain numeric string index (:issue:`28556`)
+- :meth:`read_sql` returned an empty generator if ``chunksize`` was no-zero and the query returned no results. Now returns a generator with a single empty dataframe (:issue:`34411`)
 
 Period
 ^^^^^^

pandas/io/sql.py

@@ -926,11 +926,17 @@ def _query_iterator(
         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
                 )
@@ -1343,11 +1349,21 @@ def _query_iterator(
         dtype: Optional[DtypeArg] = 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,
@@ -1849,14 +1865,20 @@ def _query_iterator(
         dtype: Optional[DtypeArg] = 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

@@ -661,6 +661,12 @@ 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):
+        query = "SELECT * FROM iris_view WHERE SepalLength < 0.0"
+        with_batch = sql.read_sql_query(query, self.conn, chunksize=5)
+        without_batch = sql.read_sql_query(query, 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)