test_Producer.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
import pytest

from confluent_kafka import Producer, Consumer, KafkaError, KafkaException, \
    TopicPartition, libversion
from struct import pack


def error_cb(err):
    print('error_cb', err)


def test_basic_api():
    """ Basic API tests, these wont really do anything since there is no
        broker configured. """

    with pytest.raises(TypeError) as ex:
        p = Producer()
    assert ex.match('expected configuration dict')

    p = Producer({'socket.timeout.ms': 10,
                  'error_cb': error_cb,
                  'message.timeout.ms': 10})

    p.produce('mytopic')
    p.produce('mytopic', value='somedata', key='a key')

    def on_delivery(err, msg):
        print('delivery', err, msg)
        # Since there is no broker, produced messages should time out.
        assert err.code() == KafkaError._MSG_TIMED_OUT
        print('message latency', msg.latency())

    p.produce(topic='another_topic', value='testing', partition=9,
              callback=on_delivery)

    p.poll(0.001)

    p.flush(0.002)
    p.flush()

    try:
        p.list_topics(timeout=0.2)
    except KafkaException as e:
        assert e.args[0].code() in (KafkaError._TIMED_OUT, KafkaError._TRANSPORT)


def test_produce_timestamp():
    """ Test produce() with timestamp arg """
    p = Producer({'socket.timeout.ms': 10,
                  'error_cb': error_cb,
                  'message.timeout.ms': 10})

    # Requires librdkafka >=v0.9.4

    try:
        p.produce('mytopic', timestamp=1234567)
    except NotImplementedError:
        # Should only fail on non-supporting librdkafka
        if libversion()[1] >= 0x00090400:
            raise

    p.flush()


# Should be updated to 0.11.4 when it is released
@pytest.mark.skipif(libversion()[1] < 0x000b0400,
                    reason="requires librdkafka >=0.11.4")
def test_produce_headers():
    """ Test produce() with timestamp arg """
    p = Producer({'socket.timeout.ms': 10,
                  'error_cb': error_cb,
                  'message.timeout.ms': 10})

    binval = pack('hhl', 1, 2, 3)

    headers_to_test = [
        [('headerkey', 'headervalue')],
        [('dupkey', 'dupvalue'), ('empty', ''), ('dupkey', 'dupvalue')],
        [('dupkey', 'dupvalue'), ('dupkey', 'diffvalue')],
        [('key_with_null_value', None)],
        [('binaryval', binval)],
        [('alreadyutf8', u'Småland'.encode('utf-8'))],
        [('isunicode', 'Jämtland')],

        {'headerkey': 'headervalue'},
        {'dupkey': 'dupvalue', 'empty': '', 'dupkey': 'dupvalue'},  # noqa: F601
        {'dupkey': 'dupvalue', 'dupkey': 'diffvalue'},  # noqa: F601
        {'key_with_null_value': None},
        {'binaryval': binval},
        {'alreadyutf8': u'Småland'.encode('utf-8')},
        {'isunicode': 'Jämtland'}
        ]

    for headers in headers_to_test:
        print('headers', type(headers), headers)
        p.produce('mytopic', value='somedata', key='a key', headers=headers)
        p.produce('mytopic', value='somedata', headers=headers)

    with pytest.raises(TypeError):
        p.produce('mytopic', value='somedata', key='a key', headers=('a', 'b'))

    with pytest.raises(TypeError):
        p.produce('mytopic', value='somedata', key='a key', headers=[('malformed_header')])

    with pytest.raises(TypeError):
        p.produce('mytopic', value='somedata', headers={'anint': 1234})

    p.flush()


# Should be updated to 0.11.4 when it is released
@pytest.mark.skipif(libversion()[1] >= 0x000b0400,
                    reason="Old versions should fail when using headers")
def test_produce_headers_should_fail():
    """ Test produce() with timestamp arg """
    p = Producer({'socket.timeout.ms': 10,
                  'error_cb': error_cb,
                  'message.timeout.ms': 10})

    with pytest.raises(NotImplementedError) as ex:
        p.produce('mytopic', value='somedata', key='a key', headers=[('headerkey', 'headervalue')])
    assert ex.match('Producer message headers requires confluent-kafka-python built for librdkafka version >=v0.11.4')


def test_subclassing():
    class SubProducer(Producer):
        def __init__(self, conf, topic):
            super(SubProducer, self).__init__(conf)
            self.topic = topic

        def produce_hi(self):
            super(SubProducer, self).produce(self.topic, value='hi')

    sp = SubProducer(dict(), 'atopic')
    assert isinstance(sp, SubProducer)

    # Invalid config should fail
    with pytest.raises(KafkaException):
        sp = SubProducer({'should.fail': False}, 'mytopic')

    sp = SubProducer({'log.thread.name': True}, 'mytopic')
    sp.produce('someother', value='not hello')
    sp.produce_hi()


def test_dr_msg_errstr():
    """
    Test that the error string for failed messages works (issue #129).
    The underlying problem is that librdkafka reuses the message payload
    for error value on Consumer messages, but on Producer messages the
    payload is the original payload and no rich error string exists.
    """
    p = Producer({"message.timeout.ms": 10})

    def handle_dr(err, msg):
        # Neither message payloads must not affect the error string.
        assert err is not None
        assert err.code() == KafkaError._MSG_TIMED_OUT
        assert "Message timed out" in err.str()

    # Unicode safe string
    p.produce('mytopic', "This is the message payload", on_delivery=handle_dr)

    # Invalid unicode sequence
    p.produce('mytopic', "\xc2\xc2", on_delivery=handle_dr)

    p.flush()


def test_set_partitioner_murmur2():
    """
    Test ability to set built-in partitioner type murmur
    """
    Producer({'partitioner': 'murmur2'})


def test_set_partitioner_murmur2_random():
    """
    Test ability to set built-in partitioner type murmur2_random
    """
    Producer({'partitioner': 'murmur2_random'})


def test_set_invalid_partitioner_murmur():
    """
    Assert invalid partitioner raises KafkaException
    """
    with pytest.raises(KafkaException) as ex:
        Producer({'partitioner': 'murmur'})
    assert ex.match('Invalid value for configuration property "partitioner": murmur')


def test_transaction_api():
    """ Excercise the transactional API """
    p = Producer({"transactional.id": "test"})

    with pytest.raises(KafkaException) as ex:
        p.init_transactions(0.5)
    assert ex.value.args[0].code() == KafkaError._TIMED_OUT
    assert ex.value.args[0].retriable() is True
    assert ex.value.args[0].fatal() is False
    assert ex.value.args[0].txn_requires_abort() is False

    # Any subsequent APIs will fail since init did not succeed.
    with pytest.raises(KafkaException) as ex:
        p.begin_transaction()
    assert ex.value.args[0].code() == KafkaError._CONFLICT
    assert ex.value.args[0].retriable() is True
    assert ex.value.args[0].fatal() is False
    assert ex.value.args[0].txn_requires_abort() is False

    consumer = Consumer({"group.id": "testgroup"})
    group_metadata = consumer.consumer_group_metadata()
    consumer.close()

    with pytest.raises(KafkaException) as ex:
        p.send_offsets_to_transaction([TopicPartition("topic", 0, 123)],
                                      group_metadata)
    assert ex.value.args[0].code() == KafkaError._CONFLICT
    assert ex.value.args[0].retriable() is True
    assert ex.value.args[0].fatal() is False
    assert ex.value.args[0].txn_requires_abort() is False

    with pytest.raises(KafkaException) as ex:
        p.commit_transaction(0.5)
    assert ex.value.args[0].code() == KafkaError._CONFLICT
    assert ex.value.args[0].retriable() is True
    assert ex.value.args[0].fatal() is False
    assert ex.value.args[0].txn_requires_abort() is False

    with pytest.raises(KafkaException) as ex:
        p.abort_transaction(0.5)
    assert ex.value.args[0].code() == KafkaError._CONFLICT
    assert ex.value.args[0].retriable() is True
    assert ex.value.args[0].fatal() is False
    assert ex.value.args[0].txn_requires_abort() is False


def test_purge():
    """
    Verify that when we have a higher message.timeout.ms timeout, we can use purge()
    to stop waiting for messages and get delivery reports
    """
    p = Producer(
        {"socket.timeout.ms": 10, "error_cb": error_cb, "message.timeout.ms": 30000}
    )  # 30 seconds

    # Hack to detect on_delivery was called because inner functions can modify nonlocal objects.
    # When python2 support is dropped, we can use the "nonlocal" keyword instead
    cb_detector = {"on_delivery_called": False}

    def on_delivery(err, msg):
        cb_detector["on_delivery_called"] = True
        # Because we are purging messages, we should see a PURGE_QUEUE kafka error
        assert err.code() == KafkaError._PURGE_QUEUE

    # Our message won't be delivered, but also won't timeout yet because our timeout is 30s.
    p.produce(topic="some_topic", value="testing", partition=9, callback=on_delivery)
    p.flush(0.002)
    assert not cb_detector["on_delivery_called"]

    # When in_queue set to false, we won't purge the message and get delivery callback
    p.purge(in_queue=False)
    p.flush(0.002)
    assert not cb_detector["on_delivery_called"]

    # When we purge including the queue, the message should have delivered a delivery report
    # with a PURGE_QUEUE error
    p.purge()
    p.flush(0.002)
    assert cb_detector["on_delivery_called"]


def test_producer_bool_value():
    """
    Make sure producer has a truth-y bool value
    See https://github.com/confluentinc/confluent-kafka-python/issues/1427
    """

    p = Producer({})
    assert bool(p)