Exemplo n.º 1
0
    def test_read_from_other_schema(self, mock_connect, __):

        inst = Postgres(self.source, OPTIONS)
        inst.tables = [
            {'value': 'my_schema.foo_bar'},
            {'value': 'your_schema.bar_foo'}
        ]
        cursor_return_value = mock_connect.return_value.cursor.return_value
        mock_data = [self.mock_recs[:1], []] * 2
        cursor_return_value.fetchall.side_effect = mock_data

        expected = [
            ('my_schema', 'foo_bar'),
            (None, None),
            ('your_schema', 'bar_foo')
        ]

        for expected_schema, expected_table in expected:
            result = inst.read()
            if result:
                schema = result[0]['__schemaname']
                table = result[0]['__tablename']

                self.assertEqual(schema, expected_schema)
                self.assertEqual(table, expected_table)
Exemplo n.º 2
0
    def test_recover_from_state(self, mock_connect, mock_execute):
        """continues to read a table from the saved state"""

        tables = [
            {
                'value': 'public.test1'
            },
            {
                'value': 'public.test2'
            },
            {
                'value': 'public.test3'
            },
        ]
        last_index = 1

        self.source['state'] = {'last_index': last_index}
        inst = Postgres(self.source, OPTIONS)
        inst.tables = tables
        cursor_return_value = mock_connect.return_value.cursor.return_value
        cursor_return_value.fetchall.return_value = [{
            'id': 101
        }, {
            'id': 102
        }, {
            'id': 103
        }]

        inst.read()
        first_query = mock_execute.call_args_list[0][0][0]
        self.assertTrue("\"id\" >= '1'" in first_query)
        self.assertTrue('FROM "public"."test2"' in first_query)
Exemplo n.º 3
0
    def test_read_retries(self, mock_connect):
        inst = Postgres(self.source, OPTIONS)
        inst.tables = [{'value': 'my_schema.foo_bar'}]
        mock_connect.side_effect = psycopg2.DatabaseError('TestRetriesError')
        with self.assertRaises(psycopg2.DatabaseError):
            inst.read()

        self.assertEqual(mock_connect.call_count, MAX_RETRIES)
Exemplo n.º 4
0
 def test_invalid_inckey(self, mock_connect):
     tables = [{'value': 'public.table1'}]
     inst = Postgres(self.source, OPTIONS)
     inst.tables = tables
     mock_connector = mock.Mock()
     mock_connector.cursor.execute.side_effect = \
         psycopg2.errors.UndefinedColumn('column does not exist')
     inst.connector = mock_connector
     with self.assertRaises(PostgresInckeyError):
         inst.read()
Exemplo n.º 5
0
    def test_incremental(self, mock_connect):
        inst = Postgres(self.source, OPTIONS)
        inst.tables = [{'value': 'schema.foo'}]
        inst.read()

        q = ('DECLARE cur CURSOR FOR '
             'SELECT * FROM "schema"."foo" '
             'WHERE "id" >= \'1\' ORDER BY "id"')
        execute_mock = mock_connect.return_value.cursor.return_value.execute
        execute_mock.assert_has_calls([mock.call(q)], True)
Exemplo n.º 6
0
    def test_use_inckey_as_index(self, mock_connect):
        inst = Postgres(self.source, OPTIONS)
        inst.tables = [{'value': 'my_schema.foo_bar'}]

        cursor_return_value = mock_connect.return_value.cursor.return_value
        cursor_return_value.fetchall.return_value = [{'id': 1, 'col1': 1}]

        inst.read()

        expected = [self.source['inckey']]
        result = inst.keys
        self.assertEqual(result, expected)
Exemplo n.º 7
0
    def test_batch_size(self, mock_connect, mock_execute):
        customBatchSize = 42
        self.source['__batchSize'] = customBatchSize
        inst = Postgres(self.source, OPTIONS)
        inst.tables = [{'value': 'my_schema.foo_bar'}]

        cursor_return_value = mock_connect.return_value.cursor.return_value
        cursor_return_value.fetchall.return_value = self.mock_recs

        inst.read()
        second_query = mock_execute.call_args_list[1][0][0]
        txt = 'FETCH FORWARD {}'.format(customBatchSize)
        self.assertTrue(second_query.startswith(txt))
Exemplo n.º 8
0
    def test_read(self, mock_connect):
        """reads a table from the database and validates that each row
        has a __tablename, __schemaname and __databasename column"""

        inst = Postgres(self.source, OPTIONS)
        inst.tables = [{'value': 'my_schema.foo_bar'}]
        cursor_return_value = mock_connect.return_value.cursor.return_value
        cursor_return_value.fetchall.return_value = self.mock_recs

        rows = inst.read()
        self.assertEqual(len(rows), len(self.mock_recs))
        for x in range(0, len(rows)):
            self.assertEqual(rows[x]['__tablename'], 'foo_bar')
            self.assertEqual(rows[x]['__schemaname'], 'my_schema')
            self.assertEqual(rows[x]['__databasename'], self.source['db_name'])
Exemplo n.º 9
0
    def test_no_state_for_empty_results(self, mock_connect, mock_state):
        """before returning a batch of data, the sources state should be
        reported as well as having the state ID appended to each data object"""

        inst = Postgres(self.source, OPTIONS)
        table_name = 'my_schema.foo_bar'
        inst.tables = [{'value': table_name}]
        result_order = [[], []]
        cursor_return_value = mock_connect.return_value.cursor.return_value
        cursor_return_value.fetchall.side_effect = result_order

        inst.read()

        # State function was called with relevant table name and row count
        mock_state.assert_not_called()
Exemplo n.º 10
0
    def test_retry_cursor_from_last_value(self, mock_connect):
        inst = Postgres(self.source, OPTIONS)
        inst.tables = [{'value': 'my_schema.foo_bar'}]
        inst.inckey = None
        inst.incval = None
        inst.keys = ['col1', 'col2']
        records = [{
            'id': 1,
            'col1': 'foo1',
            'col2': 'bar1'
        }, {
            'id': 2,
            'col1': 'foo2',
            'col2': 'bar2'
        }, {
            'id': 3,
            'col1': 'foo3',
            'col2': 'bar3'
        }]

        cursor_execute = mock_connect.return_value.cursor.return_value.execute
        cursor_execute.side_effect = [
            lambda *args: None, lambda *args: None,
            psycopg2.DatabaseError('TestRetriesError'), lambda *args: None,
            lambda *args: None
        ]
        cursor_return_value = mock_connect.return_value.cursor.return_value
        cursor_return_value.fetchall.return_value = records
        # First read no error
        inst.read()

        init_query = cursor_execute.call_args_list[0][0][0]
        expected_init_query = 'DECLARE cur CURSOR FOR SELECT * FROM ' \
                              '"my_schema"."foo_bar" ORDER BY "col1", "col2"'

        # Raise retry error and restart cursor
        inst.read()

        retried_query = cursor_execute.call_args_list[-2][0][0]
        expected_retried_query = 'DECLARE cur CURSOR FOR SELECT * FROM ' \
                                 '"my_schema"."foo_bar" WHERE ' \
                                 '"col1" >= \'foo3\' AND "col2" >= \'bar3\' ' \
                                 'ORDER BY "col1", "col2"'

        self.assertEqual(init_query, expected_init_query)
        self.assertEqual(retried_query, expected_retried_query)
Exemplo n.º 11
0
    def test_reports_state(self, mock_connect, mock_state):
        """before returning a batch of data, the sources state should be
        reported as well as having the state ID appended to each data object"""

        inst = Postgres(self.source, OPTIONS)
        table_name = 'my_schema.foo_bar'
        inst.tables = [{'value': table_name}]
        result_order = [self.mock_recs, []]
        cursor_return_value = mock_connect.return_value.cursor.return_value
        cursor_return_value.fetchall.side_effect = result_order

        rows = inst.read()
        state_id = rows[0]['__state']
        state_obj = dict([
            ('last_index', 0),
        ])

        msg = 'State ID is not the same in all rows!'
        for row in rows:
            self.assertEqual(row['__state'], state_id, msg)

        # State function was called with relevant table name and row count
        mock_state.assert_called_with(state_id, state_obj)
Exemplo n.º 12
0
    def test_read_end_stream(self, mock_connect, mock_execute):
        """reads the entire table from the database and validates that the
        stream returns None to indicate the end"""
        tables = [
            {
                'value': 'public.table1'
            },
            {
                'value': 'public.table2'
            },
            {
                'value': 'public.table3'
            },
        ]

        inst = Postgres(self.source, OPTIONS)
        inst.tables = tables
        result_order = [
            self.mock_recs, [], self.mock_recs, [], self.mock_recs, []
        ]

        cursor_return_value = mock_connect.return_value.cursor.return_value
        cursor_return_value.fetchall.side_effect = result_order

        # First call to read
        result = inst.read()
        self.assertEqual(len(result), len(self.mock_recs))
        query = mock_execute.call_args_list[0][0][0]
        expected_query = 'FROM "public"."table1" ' \
                         'WHERE "id" >= \'1\' ' \
                         'ORDER BY "id"'
        self.assertTrue(expected_query in query)
        query = mock_execute.call_args_list[1][0][0]
        expected_query = 'FETCH FORWARD'
        self.assertTrue(expected_query in query)

        # Second call to read
        result = inst.read()
        self.assertEqual(result, [])
        query = mock_execute.call_args_list[2][0][0]
        expected_query = 'FETCH FORWARD'
        self.assertTrue(expected_query in query)

        # Third call to read
        result = inst.read()
        self.assertEqual(len(result), len(self.mock_recs))
        query = mock_execute.call_args_list[3][0][0]
        expected_query = 'FROM "public"."table2" ' \
                         'WHERE "id" >= \'1\' ' \
                         'ORDER BY "id"'
        self.assertTrue(expected_query in query)
        query = mock_execute.call_args_list[4][0][0]
        expected_query = 'FETCH FORWARD'
        self.assertTrue(expected_query in query)

        # Fourth call to read
        result = inst.read()
        self.assertEqual(result, [])
        query = mock_execute.call_args_list[5][0][0]
        expected_query = 'FETCH FORWARD'
        self.assertTrue(expected_query in query)

        # Fifth call to read
        result = inst.read()
        self.assertEqual(len(result), len(self.mock_recs))
        query = mock_execute.call_args_list[6][0][0]
        expected_query = 'FROM "public"."table3" ' \
                         'WHERE "id" >= \'1\' ' \
                         'ORDER BY "id"'
        self.assertTrue(expected_query in query)
        query = mock_execute.call_args_list[7][0][0]
        expected_query = 'FETCH FORWARD'
        self.assertTrue(expected_query in query)

        # Sixth call to read
        result = inst.read()
        self.assertEqual(result, [])
        query = mock_execute.call_args_list[8][0][0]
        expected_query = 'FETCH FORWARD'
        self.assertTrue(expected_query in query)

        end = inst.read()
        self.assertEqual(end, None)