feat: use DML batches in executemany() method

google/cloud/spanner_dbapi/connection.py

@@ -32,6 +32,8 @@
 from google.cloud.spanner_dbapi.version import DEFAULT_USER_AGENT
 from google.cloud.spanner_dbapi.version import PY_VERSION
 
+from google.rpc.code_pb2 import ABORTED
+
 
 AUTOCOMMIT_MODE_WARNING = "This method is non-operational in autocommit mode"
 MAX_INTERNAL_RETRIES = 50
@@ -175,25 +177,41 @@ def _rerun_previous_statements(self):
         from the last transaction.
         """
         for statement in self._statements:
-            res_iter, retried_checksum = self.run_statement(statement, retried=True)
-            # executing all the completed statements
-            if statement != self._statements[-1]:
-                for res in res_iter:
-                    retried_checksum.consume_result(res)
-
-                _compare_checksums(statement.checksum, retried_checksum)
-            # executing the failed statement
+            if isinstance(statement, list):
+                statements, checksum = statement
+
+                transaction = self.transaction_checkout()
+                status, res = transaction.batch_update(statements)
+
+                if status.code == ABORTED:
+                    self.connection._transaction = None
+                    raise Aborted(status.details)
+
+                retried_checksum = ResultsChecksum()
+                retried_checksum.consume_result(res)
+                retried_checksum.consume_result(status.code)
+
+                _compare_checksums(checksum, retried_checksum)
             else:
-                # streaming up to the failed result or
-                # to the end of the streaming iterator
-                while len(retried_checksum) < len(statement.checksum):
-                    try:
-                        res = next(iter(res_iter))
+                res_iter, retried_checksum = self.run_statement(statement, retried=True)
+                # executing all the completed statements
+                if statement != self._statements[-1]:
+                    for res in res_iter:
                         retried_checksum.consume_result(res)
-                    except StopIteration:
-                        break
 
-                _compare_checksums(statement.checksum, retried_checksum)
+                    _compare_checksums(statement.checksum, retried_checksum)
+                # executing the failed statement
+                else:
+                    # streaming up to the failed result or
+                    # to the end of the streaming iterator
+                    while len(retried_checksum) < len(statement.checksum):
+                        try:
+                            res = next(iter(res_iter))
+                            retried_checksum.consume_result(res)
+                        except StopIteration:
+                            break
+
+                    _compare_checksums(statement.checksum, retried_checksum)
 
     def transaction_checkout(self):
         """Get a Cloud Spanner transaction.

google/cloud/spanner_dbapi/cursor.py

@@ -41,6 +41,8 @@
 from google.cloud.spanner_dbapi.utils import PeekIterator
 from google.cloud.spanner_dbapi.utils import StreamedManyResultSets
 
+from google.rpc.code_pb2 import ABORTED, OK
+
 _UNSET_COUNT = -1
 
 ColumnDetails = namedtuple("column_details", ["null_ok", "spanner_type"])
@@ -156,6 +158,15 @@ def _do_execute_update(self, transaction, sql, params):
 
         return result
 
+    def _do_batch_update(self, transaction, statements, many_result_set):
+        status, res = transaction.batch_update(statements)
+        many_result_set.add_iter(res)
+
+        if status.code == ABORTED:
+            raise Aborted(status.details)
+        elif status.code != OK:
+            raise OperationalError(status.details)
+
     def execute(self, sql, args=None):
         """Prepares and executes a Spanner database operation.
 
@@ -258,9 +269,50 @@ def executemany(self, operation, seq_of_params):
 
         many_result_set = StreamedManyResultSets()
 
-        for params in seq_of_params:
-            self.execute(operation, params)
-            many_result_set.add_iter(self._itr)
+        if classification in (parse_utils.STMT_INSERT, parse_utils.STMT_UPDATING):
+            statements = []
+
+            for params in seq_of_params:
+                sql, params = parse_utils.sql_pyformat_args_to_spanner(
+                    operation, params
+                )
+                statements.append((sql, params, get_param_types(params)))
+
+            if self.connection.autocommit:
+                self.connection.database.run_in_transaction(
+                    self._do_batch_update, statements, many_result_set
+                )
+            else:
+                retried = False
+                while True:
+                    try:
+                        transaction = self.connection.transaction_checkout()
+
+                        res_checksum = ResultsChecksum()
+                        if not retried:
+                            self.connection._statements.append(
+                                (statements, res_checksum)
+                            )
+
+                        status, res = transaction.batch_update(statements)
+                        many_result_set.add_iter(res)
+                        res_checksum.consume_result(res)
+                        res_checksum.consume_result(status.code)
+
+                        if status.code == ABORTED:
+                            self.connection._transaction = None
+                            raise Aborted(status.details)
+                        elif status.code != OK:
+                            raise OperationalError(status.details)
+                        break
+                    except Aborted:
+                        self.connection.retry_transaction()
+                        retried = True
+
+        else:
+            for params in seq_of_params:
+                self.execute(operation, params)
+                many_result_set.add_iter(self._itr)
 
         self._result_set = many_result_set
         self._itr = many_result_set