fix batch serializers

CHANGES.rst

@@ -2,6 +2,14 @@
 Changelog
 =========
 
+Unreleased
+==========
+
+Bugfixes:
+
+* Fix serialization for batch (issue #886, pr #887 by @ydjin0602)
+
+
 0.10.0 (2023-12-15)
 ===================
 

aiokafka/producer/message_accumulator.py

@@ -14,8 +14,10 @@
 
 
 class BatchBuilder:
-    def __init__(self, magic, batch_size, compression_type,
-                 *, is_transactional):
+    def __init__(
+        self, magic, batch_size, compression_type,
+        *, is_transactional, key_serializer=None, value_serializer=None
+    ):
         if magic < 2:
             assert not is_transactional
             self._builder = LegacyRecordBatchBuilder(
@@ -28,6 +30,20 @@
         self._relative_offset = 0
         self._buffer = None
         self._closed = False
+        self._key_serializer = key_serializer
+        self._value_serializer = value_serializer
+
+    def _serialize(self, key, value):
+        if self._key_serializer is None:
+            serialized_key = key
+        else:
+            serialized_key = self._key_serializer(key)
+        if self._value_serializer is None:
+            serialized_value = value
+        else:
+            serialized_value = self._value_serializer(value)
+
+        return serialized_key, serialized_value
 
     def append(self, *, timestamp, key, value, headers=[]):
         """Add a message to the batch.
@@ -49,8 +65,9 @@
         if self._closed:
             return None
 
+        key_bytes, value_bytes = self._serialize(key, value)
         metadata = self._builder.append(
-            self._relative_offset, timestamp, key, value,
+            self._relative_offset, timestamp, key=key_bytes, value=value_bytes,
             headers=headers)
 
         # Check if we could add the message
@@ -422,7 +439,7 @@
 
         return nodes, unknown_leaders_exist
 
-    def create_builder(self):
+    def create_builder(self, key_serializer=None, value_serializer=None):
         if self._api_version >= (0, 11):
             magic = 2
         elif self._api_version >= (0, 10):
@@ -435,8 +452,13 @@
                 self._txn_manager.transactional_id is not None:
             is_transactional = True
         return BatchBuilder(
-            magic, self._batch_size, self._compression_type,
-            is_transactional=is_transactional)
+            magic,
+            self._batch_size,
+            self._compression_type,
+            is_transactional=is_transactional,
+            key_serializer=key_serializer,
+            value_serializer=value_serializer
+        )
 
     def _append_batch(self, builder, tp):
         # We must do this before actual add takes place to check for errors.

aiokafka/producer/producer.py

@@ -348,14 +348,14 @@
         return (await self.client._wait_on_metadata(topic))
 
     def _serialize(self, topic, key, value):
-        if self._key_serializer:
-            serialized_key = self._key_serializer(key)
-        else:
+        if self._key_serializer is None:
             serialized_key = key
-        if self._value_serializer:
-            serialized_value = self._value_serializer(value)
         else:
+            serialized_key = self._key_serializer(key)
+        if self._value_serializer is None:
             serialized_value = value
+        else:
+            serialized_value = self._value_serializer(value)
 
         message_size = LegacyRecordBatchBuilder.record_overhead(
             self._producer_magic)
@@ -484,7 +484,9 @@
         Returns:
             BatchBuilder: empty batch to be filled and submitted by the caller.
         """
-        return self._message_accumulator.create_builder()
+        return self._message_accumulator.create_builder(
+            key_serializer=self._key_serializer, value_serializer=self._value_serializer
+        )
 
     async def send_batch(self, batch, topic, *, partition):
         """Submit a BatchBuilder for publication.

tests/test_producer.py

@@ -380,6 +380,53 @@
             await producer.send_batch(
                 batch, self.topic, partition=partition)
 
+    @run_until_complete
+    async def test_producer_send_batch_with_serializer(self):
+        def key_serializer(val):
+            return val.upper().encode()
+
+        def value_serializer(val):
+            return json.dumps(val, separators=(',', ':')).encode()
+
+        producer = AIOKafkaProducer(
+            bootstrap_servers=self.hosts,
+            key_serializer=key_serializer,
+            value_serializer=value_serializer,
+        )
+        await producer.start()
+
+        partitions = await producer.partitions_for(self.topic)
+        partition = partitions.pop()
+
+        batch = producer.create_batch()
+        batch.append(key="key1", value={"value": 111}, timestamp=None)
+        batch.append(key="key2", value={"value": 222}, timestamp=None)
+        self.assertEqual(batch.record_count(), 2)
+
+        # batch gets properly sent
+        future = await producer.send_batch(
+            batch, self.topic, partition=partition)
+        resp = await future
+        await producer.stop()
+        self.assertEqual(resp.partition, partition)
+
+        consumer = AIOKafkaConsumer(
+            self.topic,
+            bootstrap_servers=self.hosts,
+            enable_auto_commit=True,
+            auto_offset_reset="earliest")
+        await consumer.start()
+
+        msg = await consumer.getone()
+        self.assertEqual(msg.key, b"KEY1")
+        self.assertEqual(msg.value, b"{\"value\":111}")
+
+        msg = await consumer.getone()
+        self.assertEqual(msg.key, b"KEY2")
+        self.assertEqual(msg.value, b"{\"value\":222}")
+
+        await consumer.stop()
+
     @pytest.mark.ssl
     @run_until_complete
     async def test_producer_ssl(self):