factor out some basic kafka helpers

2 files changed, 23 insertions, 0 deletions

diff --git a/python/fatcat_tools/__init__.py b/python/fatcat_tools/__init__.py
index f2798f0b..78c5c90b 100644
--- a/python/fatcat_tools/__init__.py
+++ b/python/fatcat_tools/__init__.py
@@ -2,3 +2,4 @@
 from .api_auth import authenticated_api, public_api
 from .fcid import fcid2uuid, uuid2fcid
 from .transforms import *
+from .kafka import simple_kafka_producer, kafka_fail_fast
diff --git a/python/fatcat_tools/kafka.py b/python/fatcat_tools/kafka.py
new file mode 100644
index 00000000..53b62a37
--- /dev/null
+++ b/python/fatcat_tools/kafka.py
@@ -0,0 +1,22 @@
+
+from confluent_kafka import Consumer, Producer, KafkaException
+
+
+def kafka_fail_fast(err, msg):
+    if err is not None:
+        print("Kafka producer delivery error: {}".format(err))
+        print("Bailing out...")
+        # TODO: should it be sys.exit(-1)?
+        raise KafkaException(err)
+
+def simple_kafka_producer(kafka_hosts):
+
+    kafka_config = {
+        'bootstrap.servers': kafka_hosts,
+        'message.max.bytes': 20000000, # ~20 MBytes; broker-side max is ~50 MBytes
+        'delivery.report.only.error': True,
+        'default.topic.config': {
+            'request.required.acks': -1,
+        },
+    }
+    return Producer(kafka_config)