def test_kafka_consumer_offsets_search_many_partitions(kafka_consumer, kafka_producer, topic):
tp0 = TopicPartition(topic, 0)
tp1 = TopicPartition(topic, 1)
send_time = int(time.time() * 1000)
timeout = 10
p0msg = kafka_producer.send(
topic, partition=0, value=b"XXX",
timestamp_ms=send_time).get(timeout)
p1msg = kafka_producer.send(
topic, partition=1, value=b"XXX",
timestamp_ms=send_time).get(timeout)
consumer = kafka_consumer
offsets = consumer.offsets_for_times({
tp0: send_time,
tp1: send_time
})
assert offsets == {
tp0: OffsetAndTimestamp(p0msg.offset, send_time),
tp1: OffsetAndTimestamp(p1msg.offset, send_time)
}
offsets = consumer.beginning_offsets([tp0, tp1])
assert offsets == {
tp0: p0msg.offset,
tp1: p1msg.offset
}
offsets = consumer.end_offsets([tp0, tp1])
assert offsets == {
tp0: p0msg.offset + 1,
tp1: p1msg.offset + 1
}
def test_kafka_consumer_offsets_search_many_partitions(self):
tp0 = TopicPartition(self.topic, 0)
tp1 = TopicPartition(self.topic, 1)
kafka_producer = self.kafka_producer()
send_time = int(time.time() * 1000)
p0msg = kafka_producer.send(
self.topic, partition=0, value=b"XXX",
timestamp_ms=send_time).get()
p1msg = kafka_producer.send(
self.topic, partition=1, value=b"XXX",
timestamp_ms=send_time).get()
consumer = self.kafka_consumer()
offsets = consumer.offsets_for_times({
tp0: send_time,
tp1: send_time
})
self.assertEqual(offsets, {
tp0: OffsetAndTimestamp(p0msg.offset, send_time),
tp1: OffsetAndTimestamp(p1msg.offset, send_time)
})
offsets = consumer.beginning_offsets([tp0, tp1])
self.assertEqual(offsets, {
tp0: p0msg.offset,
tp1: p1msg.offset
})
offsets = consumer.end_offsets([tp0, tp1])
self.assertEqual(offsets, {
tp0: p0msg.offset + 1,
tp1: p1msg.offset + 1
})
def _setup_consumer(self):
"""
prepare offset numbers etc. for reading from Topic
"""
# <WTF> https://github.com/dpkp/kafka-python/issues/601
self.available_topics = self.client.topics()
# </WTF>
# might as well use it
assert self.topic in self.available_topics
if (self.start_params is None) != (self.end_params is None):
raise ValueError("Both start and end params must be set or both must be None")
if self.start_params is None:
# setup partitions to read through
# TODO not checked with multiple partitions since inheriting from foxglove
# An offset is assigned to make repeatability (via a locking file) possible later on.
# and it's easier to terminate the fetch loop this way.
p_id = self.client.partitions_for_topic(self.topic)
topic_partitions = [TopicPartition(topic=self.topic, partition=p) for p in list(p_id)]
starts = self.client.beginning_offsets(topic_partitions)
ends = self.client.end_offsets(topic_partitions)
self.start_p_offsets = {
tp: OffsetAndTimestamp(offset=offset, timestamp=None) for tp, offset in starts.items()
}
self.end_p_offsets = {
tp: OffsetAndTimestamp(offset=offset - 1, timestamp=None) for tp, offset in ends.items()
}
else:
# TODO - this code was inherited from Foxglove and hasn't be checked through
# setup start and end partitions and offsets
# self.client.seek_to_beginning()
# datetime is only start/end implemented
assert isinstance(self.start_params, datetime) and isinstance(self.end_params, datetime)
start = int(self.start_params.timestamp() * 1000)
end = int(self.end_params.timestamp() * 1000)
partitions = self.client.partitions_for_topic(self.topic)
tx = {TopicPartition(topic=self.topic, partition=p): start for p in list(partitions)}
self.start_p_offsets = self.client.offsets_for_times(tx)
# if you give a timestamp after the last record it returns None
for tp, offset_details in self.start_p_offsets.items():
if offset_details is None:
raise ValueError("Start date outside of available messages")
tx = {TopicPartition(topic=self.topic, partition=p): end for p in list(partitions)}
self.end_p_offsets = self.client.offsets_for_times(tx)
# as above - out of range, for end offset give something useful
for tp, offset_details in self.end_p_offsets.items():
if offset_details is None:
# go to last message. I'm not 100% sure this is correct
end_offsets = self.client.end_offsets([tp])
offset = end_offsets[tp] - 1
self.end_p_offsets[tp] = OffsetAndTimestamp(offset=offset, timestamp=None)
def test_commit_for_times_atomic(self, mock_kconsumer):
partition_to_offset = {
TopicPartition("topic1", 0): None,
TopicPartition("topic2", 0): OffsetAndTimestamp(123, 123),
}
consumer_commit_for_times(mock_kconsumer, partition_to_offset, atomic=True)
assert mock_kconsumer.commit.call_count == 0
def get_offsets_by_times(self, timestamps, timeout_ms):
offsets = self._retrieve_offsets(timestamps, timeout_ms)
for tp in timestamps:
if tp not in offsets:
offsets[tp] = None
else:
offset, timestamp = offsets[tp]
offsets[tp] = OffsetAndTimestamp(offset, timestamp)
return offsets
def test_commit_for_times(self, mock_kconsumer):
timestamp = 123
topics = ["topic1", "topic2", "topic3"]
parts = [0, 1]
partition_to_offset = {
TopicPartition(topic, part): OffsetAndTimestamp(42, timestamp)
for topic in topics for part in parts
}
expected = {
TopicPartition(topic, part): OffsetAndMetadata(42, metadata=None)
for topic in topics for part in parts
}
consumer_commit_for_times(mock_kconsumer, partition_to_offset)
mock_kconsumer.commit.assert_called_once_with(expected)
def connect(self):
if self.client is None:
self.bootstrap_server, self.topic, self.start_params, self.end_params = \
self._decode_engine_url()
self.client = KafkaConsumer(bootstrap_servers=self.bootstrap_server)
# <WTF> https://github.com/dpkp/kafka-python/issues/601
self.available_topics = self.client.topics()
# </WTF>
# might as well use it
assert self.topic in self.available_topics
# setup start and end partitions and offsets
# self.client.seek_to_beginning()
# datetime is only start/end implemented
assert isinstance(self.start_params, datetime) and isinstance(self.end_params, datetime)
start = int(self.start_params.timestamp() * 1000)
end = int(self.end_params.timestamp() * 1000)
partitions = self.client.partitions_for_topic(self.topic)
tx = {TopicPartition(topic=self.topic, partition=p):start
for p in list(partitions)}
self.start_p_offsets = self.client.offsets_for_times(tx)
# if you give a timestamp after the last record it returns None
for tp, offset_details in self.start_p_offsets.items():
if offset_details is None:
raise ValueError("Start date outside of available messages")
tx = {TopicPartition(topic=self.topic, partition=p):end
for p in list(partitions)}
self.end_p_offsets = self.client.offsets_for_times(tx)
# as above - out of range, for end offset give something useful
for tp, offset_details in self.end_p_offsets.items():
if offset_details is None:
# go to last message. I'm not 100% sure this is correct
end_offsets = self.client.end_offsets([tp])
offset = end_offsets[tp]-1
self.end_p_offsets[tp] = OffsetAndTimestamp(offset=offset, timestamp=None)
def test_offset_for_times(mocker):
partitions = [kafka.TopicPartition('ut_topic', 0)]
offsets_for_times = {tp: OffsetAndTimestamp(42, -1) for tp in partitions}
positions = {tp: 747 for tp in partitions}
mock = mocker.Mock()
mock.offsets_for_times.return_value = offsets_for_times
mock.position.side_effect = lambda tp: positions.get(tp, 0)
# Uses returned offset for time when provided
offsets = client.offsets_for_times(mock, partitions, 987654321)
assert len(offsets) == len(partitions)
assert all(tp in offsets for tp in partitions)
assert offsets[partitions[0]] == 42
# When offsets_for_times returns None returns position at end
offsets_for_times[partitions[0]] = None
offsets = client.offsets_for_times(mock, partitions, 987654321)
assert len(offsets) == len(partitions)
assert all(tp in offsets for tp in partitions)
assert offsets[partitions[0]] == 747
def _answer_offsets_for_times(timestamps):
return {
tp: OffsetAndTimestamp(ts - 100, ts)
for tp, ts in timestamps.items()
}
