def __create_backups_with_part_of_referencing_same_table_in_bq():
        backup_1 = Backup(table_id='table_id_1', dataset_id='dataset_id_1')
        backup_2 = Backup(table_id='table_id_2', dataset_id='dataset_id_1')
        backup_3 = Backup(table_id='table_id_3', dataset_id='dataset_id_1')
        backup_4 = Backup(table_id='table_id_3', dataset_id='dataset_id_1')

        return [backup_4, backup_2, backup_3, backup_1]
Exemple #2
0
 def test_that_get_all_backups_sorted_will_return_only_these_with_null_deleted_column(
         self):  # nopep8 pylint: disable=C0301, W0613
     # given
     table = Table(project_id='example-proj-name',
                   dataset_id='example-dataset-name',
                   table_id='example-table-name',
                   last_checked=datetime(2017, 02, 1, 16, 30))
     table.put()
     backup1 = Backup(
         parent=table.key,
         last_modified=datetime(2017, 02, 1, 16, 30),
         created=datetime(2017, 02, 1, 16, 30),
         dataset_id='backup_dataset',
         table_id='backup1',
         numBytes=1234,
     )
     backup1.put()
     backup2 = Backup(parent=table.key,
                      last_modified=datetime(2017, 02, 1, 16, 30),
                      created=datetime(2017, 02, 1, 16, 30),
                      dataset_id='backup_dataset',
                      table_id='backup2',
                      numBytes=1234,
                      deleted=datetime(2017, 02, 10, 16, 30))
     backup2.put()
     # when
     existing_backups = Backup.get_all_backups_sorted(table.key)
     # then
     self.assertTrue(backup1 in existing_backups)
     self.assertTrue(backup2 not in existing_backups)
    def __create_valid_backups():
        backup_1 = Backup(table_id='table_id_1', dataset_id='dataset_id_1')
        backup_2 = Backup(table_id='table_id_2', dataset_id='dataset_id_1')
        backup_3 = Backup(table_id='table_id_3', dataset_id='dataset_id_1')
        backup_4 = Backup(table_id='table_id_4', dataset_id='dataset_id_1')

        return [backup_1, backup_2, backup_3, backup_4]
Exemple #4
0
    def test_should_return_the_same_parent_table_for_child_backups(self):
        # given
        table = Table(project_id='example-proj-name',
                      dataset_id='example-dataset-name',
                      table_id='example-table-name',
                      last_checked=datetime(2017, 02, 1, 16, 30))
        table.put()
        backup_one = Backup(parent=table.key,
                            last_modified=datetime(2017, 02, 1, 16, 30),
                            created=datetime(2017, 02, 1, 16, 30),
                            dataset_id='targetDatasetId',
                            table_id='targetTableId',
                            numBytes=1234)
        backup_two = Backup(parent=table.key,
                            last_modified=datetime(2018, 03, 2, 00, 00),
                            created=datetime(2018, 03, 2, 00, 00),
                            dataset_id='targetDatasetId',
                            table_id='targetTableId',
                            numBytes=1234)

        # when
        table1 = Backup.get_table(backup_one)
        table2 = Backup.get_table(backup_two)

        # then
        self.assertEqual(table, table1)
        self.assertEqual(table1, table2)
Exemple #5
0
    def test_should_not_insert_two_backup_entities_for_the_same_backup_table(
            self):  # nopep8 pylint: disable=C0301
        # given
        table = Table(project_id='example-proj-name',
                      dataset_id='example-dataset-name',
                      table_id='example-table-name',
                      last_checked=datetime(2017, 02, 1, 16, 30))
        table.put()
        backup_one = Backup(parent=table.key,
                            last_modified=datetime(2017, 02, 1, 16, 30),
                            created=datetime(2017, 02, 1, 16, 30),
                            dataset_id='targetDatasetId',
                            table_id='targetTableId',
                            numBytes=1234)
        backup_two = Backup(parent=table.key,
                            last_modified=datetime(2018, 03, 2, 00, 00),
                            created=datetime(2018, 03, 2, 00, 00),
                            dataset_id='targetDatasetId',
                            table_id='targetTableId',
                            numBytes=1234)

        # when
        Backup.insert_if_absent(backup_one)
        Backup.insert_if_absent(backup_two)
        backups = list(Backup.get_all())

        # then
        self.assertEqual(len(backups), 1)
        self.assertEqual(backup_one.created, backups[0].created)
Exemple #6
0
def create_backup(created_datetime, table=Table(), table_id=None):
    if not table_id:
        table_id = 'targetTable' + str(random.randint(1, 1000000))
    backup_size = random.randint(1, 1000)
    backup = Backup(parent=table.key,
                    created=created_datetime,
                    dataset_id='targetDataset',
                    table_id=table_id,
                    numBytes=backup_size)
    return backup
Exemple #7
0
    def test_should_return_true_if_table_was_changed_after_last_backup(self, _):
        # given
        backup = Backup(
            parent=self.table.key,
            last_modified=datetime(2016, 11, 13, 15, 00)
        )
        backup.put()
        predicate = OnDemandBackupPredicate()

        # when
        result = predicate.test(self.big_query_table_metadata, self.table)
        # then
        self.assertTrue(result, "OnDemandShouldBackupPredicate should return TRUE "
                                "if table was changed after last backup")
Exemple #8
0
    def test_should_return_true_if_table_was_changed_before_last_backup(self, _):  # nopep8 pylint: disable=C0301
        # given
        backup = Backup(
            parent=self.table.key,
            last_modified=datetime(2016, 11, 13, 15, 00)
        )
        backup.put()
        predicate = OnDemandBackupPredicate()

        # when
        result = predicate.test(self.big_query_table_metadata, self.table)
        # then
        self.assertTrue(result, "OnDemandShouldBackupPredicate should return FALSE "
                                "if table was changed before "
                                "last backup was made")
Exemple #9
0
    def test_should_return_false_if_table_was_changed_at_the_same_time_when_last_backup(
            self, _1, _2):  # nopep8 pylint: disable=C0301
        # given
        backup = Backup(parent=self.table.key,
                        last_modified=datetime(2016, 11, 13, 15, 00),
                        numBytes=123)
        backup.put()
        predicate = DefaultBackupPredicate()

        # when
        result = predicate.test(self.big_query_table_metadata, self.table)
        # then
        self.assertFalse(
            result, "ShouldBackupPredicate should return False "
            "if table was change at the same time when "
            "last backup was made")
Exemple #10
0
    def test_should_return_true_if_changed_table_is_empty_and_last_backup_is_also_empty(
            self, _1, _2):  # nopep8 pylint: disable=C0301
        # given
        backup = Backup(parent=self.table.key,
                        last_modified=datetime(2016, 11, 13, 15, 00),
                        numBytes=0)
        backup.put()
        predicate = DefaultBackupPredicate()

        # when
        result = predicate.test(self.big_query_table_metadata, self.table)

        # then
        self.assertTrue(
            result, "ShouldBackupPredicate should return True "
            "if table was changed after last backup was made,"
            "but source table is empty and bbq has also empty last backup")
Exemple #11
0
    def test_should_return_true_if_table_was_changed_before_last_backup_but_table_size_is_different(
            self, _1, _2):  # nopep8 pylint: disable=C0301
        # given
        backup = Backup(parent=self.table.key,
                        last_modified=datetime(2016, 11, 13, 15, 00),
                        numBytes=123)
        backup.put()
        predicate = DefaultBackupPredicate()

        # when
        result = predicate.test(self.big_query_table_metadata, self.table)
        # then
        self.assertTrue(
            result, "ShouldBackupPredicate should return TRUE "
            "if table was changed before "
            "last backup was made but "
            "backup has different size than source table")
Exemple #12
0
 def test_should_retrieve_table_using_backup(self):
     # given
     table = Table(project_id='example-proj-name',
                   dataset_id='example-dataset-name',
                   table_id='example-table-name',
                   last_checked=datetime(2017, 02, 1, 16, 30))
     table.put()
     backup = Backup(parent=table.key,
                     last_modified=datetime(2017, 02, 1, 16, 30),
                     created=datetime(2017, 02, 1, 16, 30),
                     dataset_id='targetDatasetId',
                     table_id='targetTableId',
                     numBytes=1234)
     backup.put()
     # then
     backup_entity = Backup.get_by_key(backup.key)
     table_entity = Table.get_table_from_backup(backup_entity)
     self.assertEqual(table_entity, table)
Exemple #13
0
 def test_that_not_deleted_backup_doesnt_have_created_field(self):
     # given
     table = Table(project_id='example-proj-name',
                   dataset_id='example-dataset-name',
                   table_id='example-table-name',
                   last_checked=datetime(2017, 02, 1, 16, 30))
     backup = Backup(parent=table.key,
                     last_modified=datetime(2017, 02, 1, 16, 30),
                     created=datetime(2017, 02, 1, 16, 30),
                     dataset_id='targetDatasetId',
                     table_id='targetTableId',
                     numBytes=1234)
     backup.put()
     # then
     backup_to_check = Backup.get_by_key(backup.key)
     self.assertEqual(backup_to_check.deleted, None)
     self.assertEqual(backup_to_check.created,
                      datetime(2017, 02, 1, 16, 30))
Exemple #14
0
 def test_deleting_backup_is_adding_current_timestamp_in_deleted_field(
         self):
     # given
     table = Table(project_id='example-proj-name',
                   dataset_id='example-dataset-name',
                   table_id='example-table-name',
                   last_checked=datetime(2017, 02, 1, 16, 30))
     backup = Backup(parent=table.key,
                     last_modified=datetime(2017, 02, 1, 16, 30),
                     created=datetime(2017, 02, 1, 16, 30),
                     dataset_id='targetDatasetId',
                     table_id='targetTableId',
                     numBytes=1234)
     backup.put()
     # when
     Backup.mark_backup_deleted(backup.key)
     # then
     deleted_backup = Backup.get_by_key(backup.key)
     self.assertEqual(deleted_backup.deleted, datetime(2017, 02, 3, 16, 30))
Exemple #15
0
    def __create_backup(backup_table_metadata, copy_job_results):

        table_entity = Table.get_table_by_reference(
            TableReference.from_bq_table(copy_job_results.source_bq_table))

        if table_entity is None:
            raise DatastoreTableGetRetriableException()

        backup = Backup(parent=table_entity.key,
                        last_modified=copy_job_results.start_time,
                        created=copy_job_results.end_time,
                        dataset_id=copy_job_results.target_dataset_id,
                        table_id=copy_job_results.target_table_id,
                        numBytes=backup_table_metadata.table_size_in_bytes())
        logging.debug(
            "Saving backup to datastore, source:{0}, target:{1}".format(
                copy_job_results.source_bq_table,
                copy_job_results.target_bq_table))

        backup.insert_if_absent(backup)
Exemple #16
0
 def __create_backup_entity(source_table_entity, dataset_id, table_id):
     return Backup(dataset_id=dataset_id,
                   table_id=table_id,
                   parent=source_table_entity.key)