kafka-python 2.2.2__tar.gz → 2.2.3__tar.gz

{kafka_python-2.2.2 → kafka_python-2.2.3}/CHANGES.md

@@ -1,3 +1,9 @@
+# 2.2.3 (May 1, 2025)
+
+Fixes
+* Ignore leading SECURITY_PROTOCOL:// in bootstrap_servers (#2608)
+* Only create fetch requests for ready nodes (#2607)
+
 # 2.2.2 (Apr 30, 2025)
 
 Fixes

{kafka_python-2.2.2 → kafka_python-2.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: kafka-python
-Version: 2.2.2
+Version: 2.2.3
 Summary: Pure Python client for Apache Kafka
 Author-email: Dana Powers <dana.powers@gmail.com>
 Project-URL: Homepage, https://github.com/dpkp/kafka-python

{kafka_python-2.2.2 → kafka_python-2.2.3}/kafka/cluster.py

@@ -3,13 +3,15 @@ from __future__ import absolute_import
 import collections
 import copy
 import logging
+import random
+import re
 import threading
 import time
 
 from kafka.vendor import six
 
 from kafka import errors as Errors
-from kafka.conn import collect_hosts
+from kafka.conn import get_ip_port_afi
 from kafka.future import Future
 from kafka.structs import BrokerMetadata, PartitionMetadata, TopicPartition
 
@@ -422,3 +424,24 @@ class ClusterMetadata(object):
     def __str__(self):
         return 'ClusterMetadata(brokers: %d, topics: %d, coordinators: %d)' % \
                (len(self._brokers), len(self._partitions), len(self._coordinators))
+
+
+def collect_hosts(hosts, randomize=True):
+    """
+    Collects a comma-separated set of hosts (host:port) and optionally
+    randomize the returned list.
+    """
+
+    if isinstance(hosts, six.string_types):
+        hosts = hosts.strip().split(',')
+
+    result = []
+    for host_port in hosts:
+        # ignore leading SECURITY_PROTOCOL:// to mimic java client
+        host_port = re.sub('^.*://', '', host_port)
+        host, port, afi = get_ip_port_afi(host_port)
+        result.append((host, port, afi))
+
+    if randomize:
+        random.shuffle(result)
+    return result

{kafka_python-2.2.2 → kafka_python-2.2.3}/kafka/conn.py

@@ -4,7 +4,7 @@ import copy
 import errno
 import io
 import logging
-from random import shuffle, uniform
+from random import uniform
 
 # selectors in stdlib as of py3.4
 try:
@@ -1496,32 +1496,6 @@ def get_ip_port_afi(host_and_port_str):
             return host, port, af
 
 
-def collect_hosts(hosts, randomize=True):
-    """
-    Collects a comma-separated set of hosts (host:port) and optionally
-    randomize the returned list.
-    """
-
-    if isinstance(hosts, six.string_types):
-        hosts = hosts.strip().split(',')
-
-    result = []
-    afi = socket.AF_INET
-    for host_port in hosts:
-
-        host, port, afi = get_ip_port_afi(host_port)
-
-        if port < 0:
-            port = DEFAULT_KAFKA_PORT
-
-        result.append((host, port, afi))
-
-    if randomize:
-        shuffle(result)
-
-    return result
-
-
 def is_inet_4_or_6(gai):
     """Given a getaddrinfo struct, return True iff ipv4 or ipv6"""
     return gai[0] in (socket.AF_INET, socket.AF_INET6)

{kafka_python-2.2.2 → kafka_python-2.2.3}/kafka/consumer/fetcher.py

@@ -153,6 +153,7 @@ class Fetcher(six.Iterator):
             future = self._client.send(node_id, request, wakeup=False)
             future.add_callback(self._handle_fetch_response, node_id, fetch_offsets, time.time())
             future.add_errback(self._handle_fetch_error, node_id)
+            future.add_both(self._clear_pending_fetch_request, node_id)
             futures.append(future)
         self._fetch_futures.extend(futures)
         self._clean_done_fetch_futures()
@@ -643,36 +644,42 @@ class Fetcher(six.Iterator):
                 log.debug("Skipping fetch for partition %s because node %s is throttled",
                         partition, node_id)
 
+            elif not self._client.ready(node_id):
+                # Until we support send request queues, any attempt to send to a not-ready node will be
+                # immediately failed with NodeNotReadyError.
+                log.debug("Skipping fetch for partition %s because connection to leader node is not ready yet")
+
             elif node_id in self._nodes_with_pending_fetch_requests:
                 log.debug("Skipping fetch for partition %s because there is a pending fetch request to node %s",
                         partition, node_id)
-                continue
 
-            if version < 5:
-                partition_info = (
-                    partition.partition,
-                    position.offset,
-                    self.config['max_partition_fetch_bytes']
-                )
-            elif version <= 8:
-                partition_info = (
-                    partition.partition,
-                    position.offset,
-                    -1, # log_start_offset is used internally by brokers / replicas only
-                    self.config['max_partition_fetch_bytes'],
-                )
             else:
-                partition_info = (
-                    partition.partition,
-                    position.leader_epoch,
-                    position.offset,
-                    -1, # log_start_offset is used internally by brokers / replicas only
-                    self.config['max_partition_fetch_bytes'],
-                )
-
-            fetchable[node_id][partition] = partition_info
-            log.debug("Adding fetch request for partition %s at offset %d",
-                      partition, position.offset)
+                # Leader is connected and does not have a pending fetch request
+                if version < 5:
+                    partition_info = (
+                        partition.partition,
+                        position.offset,
+                        self.config['max_partition_fetch_bytes']
+                    )
+                elif version <= 8:
+                    partition_info = (
+                        partition.partition,
+                        position.offset,
+                        -1, # log_start_offset is used internally by brokers / replicas only
+                        self.config['max_partition_fetch_bytes'],
+                    )
+                else:
+                    partition_info = (
+                        partition.partition,
+                        position.leader_epoch,
+                        position.offset,
+                        -1, # log_start_offset is used internally by brokers / replicas only
+                        self.config['max_partition_fetch_bytes'],
+                    )
+
+                fetchable[node_id][partition] = partition_info
+                log.debug("Adding fetch request for partition %s at offset %d",
+                          partition, position.offset)
 
         requests = {}
         for node_id, next_partitions in six.iteritems(fetchable):
@@ -761,14 +768,18 @@ class Fetcher(six.Iterator):
 
         if self._sensors:
             self._sensors.fetch_latency.record((time.time() - send_time) * 1000)
-        self._nodes_with_pending_fetch_requests.remove(node_id)
 
     def _handle_fetch_error(self, node_id, exception):
         level = logging.INFO if isinstance(exception, Errors.Cancelled) else logging.ERROR
         log.log(level, 'Fetch to node %s failed: %s', node_id, exception)
         if node_id in self._session_handlers:
             self._session_handlers[node_id].handle_error(exception)
-        self._nodes_with_pending_fetch_requests.remove(node_id)
+
+    def _clear_pending_fetch_request(self, node_id, _):
+        try:
+            self._nodes_with_pending_fetch_requests.remove(node_id)
+        except KeyError:
+            pass
 
     def _parse_fetched_data(self, completed_fetch):
         tp = completed_fetch.topic_partition

kafka_python-2.2.3/kafka/version.py

@@ -0,0 +1 @@
+__version__ = '2.2.3'

{kafka_python-2.2.2 → kafka_python-2.2.3}/kafka_python.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: kafka-python
-Version: 2.2.2
+Version: 2.2.3
 Summary: Pure Python client for Apache Kafka
 Author-email: Dana Powers <dana.powers@gmail.com>
 Project-URL: Homepage, https://github.com/dpkp/kafka-python

{kafka_python-2.2.2 → kafka_python-2.2.3}/test/test_cluster.py

@@ -1,7 +1,9 @@
 # pylint: skip-file
 from __future__ import absolute_import
 
-from kafka.cluster import ClusterMetadata
+import socket
+
+from kafka.cluster import ClusterMetadata, collect_hosts
 from kafka.protocol.metadata import MetadataResponse
 
 
@@ -132,3 +134,60 @@ def test_metadata_v7():
     assert cluster.cluster_id == 'cluster-foo'
     assert cluster._partitions['topic-1'][0].offline_replicas == [12]
     assert cluster._partitions['topic-1'][0].leader_epoch == 0
+
+
+def test_collect_hosts__happy_path():
+    hosts = "127.0.0.1:1234,127.0.0.1"
+    results = collect_hosts(hosts)
+    assert set(results) == set([
+        ('127.0.0.1', 1234, socket.AF_INET),
+        ('127.0.0.1', 9092, socket.AF_INET),
+    ])
+
+
+def test_collect_hosts__ipv6():
+    hosts = "[localhost]:1234,[2001:1000:2000::1],[2001:1000:2000::1]:1234"
+    results = collect_hosts(hosts)
+    assert set(results) == set([
+        ('localhost', 1234, socket.AF_INET6),
+        ('2001:1000:2000::1', 9092, socket.AF_INET6),
+        ('2001:1000:2000::1', 1234, socket.AF_INET6),
+    ])
+
+
+def test_collect_hosts__string_list():
+    hosts = [
+        'localhost:1234',
+        'localhost',
+        '[localhost]',
+        '2001::1',
+        '[2001::1]',
+        '[2001::1]:1234',
+    ]
+    results = collect_hosts(hosts)
+    assert set(results) == set([
+        ('localhost', 1234, socket.AF_UNSPEC),
+        ('localhost', 9092, socket.AF_UNSPEC),
+        ('localhost', 9092, socket.AF_INET6),
+        ('2001::1', 9092, socket.AF_INET6),
+        ('2001::1', 9092, socket.AF_INET6),
+        ('2001::1', 1234, socket.AF_INET6),
+    ])
+
+
+def test_collect_hosts__with_spaces():
+    hosts = "localhost:1234, localhost"
+    results = collect_hosts(hosts)
+    assert set(results) == set([
+        ('localhost', 1234, socket.AF_UNSPEC),
+        ('localhost', 9092, socket.AF_UNSPEC),
+    ])
+
+
+def test_collect_hosts__protocol():
+    hosts = "SASL_SSL://foo.bar:1234,SASL_SSL://fizz.buzz:5678"
+    results = collect_hosts(hosts)
+    assert set(results) == set([
+        ('foo.bar', 1234, socket.AF_UNSPEC),
+        ('fizz.buzz', 5678, socket.AF_UNSPEC),
+    ])

{kafka_python-2.2.2 → kafka_python-2.2.3}/test/test_conn.py

@@ -10,7 +10,7 @@ except ImportError:
     import mock
 import pytest
 
-from kafka.conn import BrokerConnection, ConnectionStates, collect_hosts
+from kafka.conn import BrokerConnection, ConnectionStates
 from kafka.future import Future
 from kafka.protocol.api import RequestHeader
 from kafka.protocol.group import HeartbeatResponse
@@ -280,54 +280,6 @@ def test_close(conn):
     pass # TODO
 
 
-def test_collect_hosts__happy_path():
-    hosts = "127.0.0.1:1234,127.0.0.1"
-    results = collect_hosts(hosts)
-    assert set(results) == set([
-        ('127.0.0.1', 1234, socket.AF_INET),
-        ('127.0.0.1', 9092, socket.AF_INET),
-    ])
-
-
-def test_collect_hosts__ipv6():
-    hosts = "[localhost]:1234,[2001:1000:2000::1],[2001:1000:2000::1]:1234"
-    results = collect_hosts(hosts)
-    assert set(results) == set([
-        ('localhost', 1234, socket.AF_INET6),
-        ('2001:1000:2000::1', 9092, socket.AF_INET6),
-        ('2001:1000:2000::1', 1234, socket.AF_INET6),
-    ])
-
-
-def test_collect_hosts__string_list():
-    hosts = [
-        'localhost:1234',
-        'localhost',
-        '[localhost]',
-        '2001::1',
-        '[2001::1]',
-        '[2001::1]:1234',
-    ]
-    results = collect_hosts(hosts)
-    assert set(results) == set([
-        ('localhost', 1234, socket.AF_UNSPEC),
-        ('localhost', 9092, socket.AF_UNSPEC),
-        ('localhost', 9092, socket.AF_INET6),
-        ('2001::1', 9092, socket.AF_INET6),
-        ('2001::1', 9092, socket.AF_INET6),
-        ('2001::1', 1234, socket.AF_INET6),
-    ])
-
-
-def test_collect_hosts__with_spaces():
-    hosts = "localhost:1234, localhost"
-    results = collect_hosts(hosts)
-    assert set(results) == set([
-        ('localhost', 1234, socket.AF_UNSPEC),
-        ('localhost', 9092, socket.AF_UNSPEC),
-    ])
-
-
 def test_lookup_on_connect():
     hostname = 'example.org'
     port = 9092

{kafka_python-2.2.2 → kafka_python-2.2.3}/test/test_fetcher.py

@@ -103,6 +103,7 @@ def test_create_fetch_requests(fetcher, mocker, api_version, fetch_version):
     fetcher._client._api_versions = BROKER_API_VERSIONS[api_version]
     mocker.patch.object(fetcher._client.cluster, "leader_for_partition", return_value=0)
     mocker.patch.object(fetcher._client.cluster, "leader_epoch_for_partition", return_value=0)
+    mocker.patch.object(fetcher._client, "ready", return_value=True)
     by_node = fetcher._create_fetch_requests()
     requests_and_offsets = by_node.values()
     assert set([r.API_VERSION for (r, _offsets) in requests_and_offsets]) == set([fetch_version])

kafka_python-2.2.2/kafka/version.py

@@ -1 +0,0 @@
-__version__ = '2.2.2'