Add 'codec' parameter to Producer

Adds a codec parameter to Producer.__init__ that lets the user choose
a compression codec to use for all messages sent by it.

Author: Patrick Lucas

kafka/producer.py

@@ -10,7 +10,10 @@
 
 from kafka.common import ProduceRequest, TopicAndPartition
 from kafka.partitioner import HashedPartitioner
-from kafka.protocol import create_message
+from kafka.protocol import (
+    CODEC_NONE, CODEC_GZIP, CODEC_SNAPPY, ALL_CODECS,
+    create_message, create_gzip_message, create_snappy_message,
+)
 
 log = logging.getLogger("kafka")
 
@@ -20,7 +23,7 @@
 STOP_ASYNC_PRODUCER = -1
 
 
-def _send_upstream(queue, client, batch_time, batch_size,
+def _send_upstream(queue, client, codec, batch_time, batch_size,
                    req_acks, ack_timeout):
     """
     Listen on the queue for a specified number of messages or till
@@ -61,7 +64,14 @@ def _send_upstream(queue, client, batch_time, batch_size,
 
         # Send collected requests upstream
         reqs = []
-        for topic_partition, messages in msgset.items():
+        for topic_partition, msg in msgset.items():
+            if codec == CODEC_GZIP:
+                messages = [create_gzip_message(msg)]
+            elif codec == CODEC_SNAPPY:
+                messages = [create_snappy_message(msg)]
+            else:
+                messages = [create_message(m) for m in msg]
+
             req = ProduceRequest(topic_partition.topic,
                                  topic_partition.partition,
                                  messages)
@@ -101,6 +111,7 @@ class Producer(object):
     def __init__(self, client, async=False,
                  req_acks=ACK_AFTER_LOCAL_WRITE,
                  ack_timeout=DEFAULT_ACK_TIMEOUT,
+                 codec=None,
                  batch_send=False,
                  batch_send_every_n=BATCH_SEND_MSG_COUNT,
                  batch_send_every_t=BATCH_SEND_DEFAULT_INTERVAL):
@@ -118,11 +129,17 @@ def __init__(self, client, async=False,
         self.req_acks = req_acks
         self.ack_timeout = ack_timeout
 
+        if codec is None:
+            codec = CODEC_NONE
+        assert codec in ALL_CODECS
+        self.codec = codec
+
         if self.async:
             self.queue = Queue()  # Messages are sent through this queue
             self.proc = Process(target=_send_upstream,
                                 args=(self.queue,
                                       self.client.copy(),
+                                      self.codec,
                                       batch_send_every_t,
                                       batch_send_every_n,
                                       self.req_acks,
@@ -138,11 +155,16 @@ def send_messages(self, topic, partition, *msg):
         """
         if self.async:
             for m in msg:
-                self.queue.put((TopicAndPartition(topic, partition),
-                                create_message(m)))
+                self.queue.put((TopicAndPartition(topic, partition), m))
             resp = []
         else:
-            messages = [create_message(m) for m in msg]
+            if self.codec == CODEC_GZIP:
+                messages = [create_gzip_message(msg)]
+            elif self.codec == CODEC_SNAPPY:
+                messages = [create_snappy_message(msg)]
+            else:
+                messages = [create_message(m) for m in msg]
+
             req = ProduceRequest(topic, partition, messages)
             try:
                 resp = self.client.send_produce_request([req], acks=self.req_acks,
@@ -167,7 +189,7 @@ def stop(self, timeout=1):
 
 class SimpleProducer(Producer):
     """
-    A simple, round-robbin producer. Each message goes to exactly one partition
+    A simple, round-robin producer. Each message goes to exactly one partition
 
     Params:
     client - The Kafka client instance to use
@@ -184,12 +206,13 @@ class SimpleProducer(Producer):
     def __init__(self, client, async=False,
                  req_acks=Producer.ACK_AFTER_LOCAL_WRITE,
                  ack_timeout=Producer.DEFAULT_ACK_TIMEOUT,
+                 codec=None,
                  batch_send=False,
                  batch_send_every_n=BATCH_SEND_MSG_COUNT,
                  batch_send_every_t=BATCH_SEND_DEFAULT_INTERVAL):
         self.partition_cycles = {}
         super(SimpleProducer, self).__init__(client, async, req_acks,
-                                             ack_timeout, batch_send,
+                                             ack_timeout, codec, batch_send,
                                              batch_send_every_n,
                                              batch_send_every_t)
 
@@ -227,6 +250,7 @@ class KeyedProducer(Producer):
     def __init__(self, client, partitioner=None, async=False,
                  req_acks=Producer.ACK_AFTER_LOCAL_WRITE,
                  ack_timeout=Producer.DEFAULT_ACK_TIMEOUT,
+                 codec=None,
                  batch_send=False,
                  batch_send_every_n=BATCH_SEND_MSG_COUNT,
                  batch_send_every_t=BATCH_SEND_DEFAULT_INTERVAL):
@@ -236,7 +260,7 @@ def __init__(self, client, partitioner=None, async=False,
         self.partitioners = {}
 
         super(KeyedProducer, self).__init__(client, async, req_acks,
-                                            ack_timeout, batch_send,
+                                            ack_timeout, codec, batch_send,
                                             batch_send_every_n,
                                             batch_send_every_t)
 

kafka/protocol.py

@@ -18,6 +18,12 @@
 
 log = logging.getLogger("kafka")
 
+ATTRIBUTE_CODEC_MASK = 0x03
+CODEC_NONE = 0x00
+CODEC_GZIP = 0x01
+CODEC_SNAPPY = 0x02
+ALL_CODECS = (CODEC_NONE, CODEC_GZIP, CODEC_SNAPPY)
+
 
 class KafkaProtocol(object):
     """
@@ -32,11 +38,6 @@ class KafkaProtocol(object):
     OFFSET_COMMIT_KEY = 8
     OFFSET_FETCH_KEY = 9
 
-    ATTRIBUTE_CODEC_MASK = 0x03
-    CODEC_NONE = 0x00
-    CODEC_GZIP = 0x01
-    CODEC_SNAPPY = 0x02
-
     ###################
     #   Private API   #
     ###################
@@ -151,17 +152,17 @@ def _decode_message(cls, data, offset):
         (key, cur) = read_int_string(data, cur)
         (value, cur) = read_int_string(data, cur)
 
-        codec = att & KafkaProtocol.ATTRIBUTE_CODEC_MASK
+        codec = att & ATTRIBUTE_CODEC_MASK
 
-        if codec == KafkaProtocol.CODEC_NONE:
+        if codec == CODEC_NONE:
             yield (offset, Message(magic, att, key, value))
 
-        elif codec == KafkaProtocol.CODEC_GZIP:
+        elif codec == CODEC_GZIP:
             gz = gzip_decode(value)
             for (offset, msg) in KafkaProtocol._decode_message_set_iter(gz):
                 yield (offset, msg)
 
-        elif codec == KafkaProtocol.CODEC_SNAPPY:
+        elif codec == CODEC_SNAPPY:
             snp = snappy_decode(value)
             for (offset, msg) in KafkaProtocol._decode_message_set_iter(snp):
                 yield (offset, msg)
@@ -544,7 +545,7 @@ def create_gzip_message(payloads, key=None):
         [create_message(payload) for payload in payloads])
 
     gzipped = gzip_encode(message_set)
-    codec = KafkaProtocol.ATTRIBUTE_CODEC_MASK & KafkaProtocol.CODEC_GZIP
+    codec = ATTRIBUTE_CODEC_MASK & CODEC_GZIP
 
     return Message(0, 0x00 | codec, key, gzipped)
 
@@ -565,6 +566,6 @@ def create_snappy_message(payloads, key=None):
         [create_message(payload) for payload in payloads])
 
     snapped = snappy_encode(message_set)
-    codec = KafkaProtocol.ATTRIBUTE_CODEC_MASK & KafkaProtocol.CODEC_SNAPPY
+    codec = ATTRIBUTE_CODEC_MASK & CODEC_SNAPPY
 
     return Message(0, 0x00 | codec, key, snapped)

test/test_unit.py

@@ -135,8 +135,8 @@ def test_create_gzip(self):
         payloads = ["v1", "v2"]
         msg = create_gzip_message(payloads)
         self.assertEqual(msg.magic, 0)
-        self.assertEqual(msg.attributes, KafkaProtocol.ATTRIBUTE_CODEC_MASK &
-                                         KafkaProtocol.CODEC_GZIP)
+        self.assertEqual(msg.attributes, ATTRIBUTE_CODEC_MASK &
+                                         CODEC_GZIP)
         self.assertEqual(msg.key, None)
         # Need to decode to check since gzipped payload is non-deterministic
         decoded = gzip_decode(msg.value)
@@ -151,8 +151,8 @@ def test_create_snappy(self):
         payloads = ["v1", "v2"]
         msg = create_snappy_message(payloads)
         self.assertEqual(msg.magic, 0)
-        self.assertEqual(msg.attributes, KafkaProtocol.ATTRIBUTE_CODEC_MASK &
-                                         KafkaProtocol.CODEC_SNAPPY)
+        self.assertEqual(msg.attributes, ATTRIBUTE_CODEC_MASK &
+                                         CODEC_SNAPPY)
         self.assertEqual(msg.key, None)
         expect = ("8\x00\x00\x19\x01@\x10L\x9f[\xc2\x00\x00\xff\xff\xff\xff"
                   "\x00\x00\x00\x02v1\x19\x1bD\x00\x10\xd5\x96\nx\x00\x00\xff"