Expose record headers in ConsumerRecords

Author: hnousiainen

Diff for: README.rst

@@ -70,6 +70,11 @@ that expose basic message attributes: topic, partition, offset, key, and value:
 >>> for msg in consumer:
 ...     assert isinstance(msg.value, dict)
 
+>>> # Access record headers. The returned value is a list of tuples
+>>> # with str, bytes for key and value
+>>> for msg in consumer:
+...     print (msg.headers)
+
 >>> # Get consumer metrics
 >>> metrics = consumer.metrics()
 

Diff for: kafka/consumer/fetcher.py

@@ -29,7 +29,7 @@
 
 ConsumerRecord = collections.namedtuple("ConsumerRecord",
     ["topic", "partition", "offset", "timestamp", "timestamp_type",
-     "key", "value", "checksum", "serialized_key_size", "serialized_value_size"])
+     "key", "value", "headers", "checksum", "serialized_key_size", "serialized_value_size", "serialized_header_size"])
 
 
 CompletedFetch = collections.namedtuple("CompletedFetch",
@@ -456,10 +456,12 @@ def _unpack_message_set(self, tp, records):
                     value = self._deserialize(
                         self.config['value_deserializer'],
                         tp.topic, record.value)
+                    headers = record.headers
+                    header_size = sum(len(h_key.encode("utf-8")) + len(h_val) for h_key, h_val in headers) if headers else -1
                     yield ConsumerRecord(
                         tp.topic, tp.partition, record.offset, record.timestamp,
-                        record.timestamp_type, key, value, record.checksum,
-                        key_size, value_size)
+                        record.timestamp_type, key, value, headers, record.checksum,
+                        key_size, value_size, header_size)
 
                 batch = records.next_batch()
 

Diff for: test/test_fetcher.py

@@ -509,7 +509,7 @@ def test_partition_records_offset():
     fetch_offset = 123
     tp = TopicPartition('foo', 0)
     messages = [ConsumerRecord(tp.topic, tp.partition, i,
-                               None, None, 'key', 'value', 'checksum', 0, 0)
+                               None, None, 'key', 'value', [], 'checksum', 0, 0, -1)
                 for i in range(batch_start, batch_end)]
     records = Fetcher.PartitionRecords(fetch_offset, None, messages)
     assert len(records) > 0
@@ -534,7 +534,7 @@ def test_partition_records_no_fetch_offset():
     fetch_offset = 123
     tp = TopicPartition('foo', 0)
     messages = [ConsumerRecord(tp.topic, tp.partition, i,
-                               None, None, 'key', 'value', 'checksum', 0, 0)
+                               None, None, 'key', 'value', None, 'checksum', 0, 0, -1)
                 for i in range(batch_start, batch_end)]
     records = Fetcher.PartitionRecords(fetch_offset, None, messages)
     assert len(records) == 0
@@ -549,7 +549,7 @@ def test_partition_records_compacted_offset():
     fetch_offset = 42
     tp = TopicPartition('foo', 0)
     messages = [ConsumerRecord(tp.topic, tp.partition, i,
-                               None, None, 'key', 'value', 'checksum', 0, 0)
+                               None, None, 'key', 'value', None, 'checksum', 0, 0, -1)
                 for i in range(batch_start, batch_end) if i != fetch_offset]
     records = Fetcher.PartitionRecords(fetch_offset, None, messages)
     assert len(records) == batch_end - fetch_offset - 1