Merge branch 'bnewbold-ingest-oa-container' into 'master'

container-ingest tool

See merge request webgroup/fatcat!8

3 files changed, 28 insertions, 2 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/importers/ingest.py b/python/fatcat_tools/importers/ingest.py
index 7dad13ce..deb4ef51 100644
--- a/python/fatcat_tools/importers/ingest.py
+++ b/python/fatcat_tools/importers/ingest.py
@@ -28,13 +28,16 @@ class IngestFileResultImporter(EntityImporter):
             print("Requiring GROBID status == 200")
         else:
             print("NOT checking GROBID success")
-        self.ingest_request_source_whitelist = ['fatcat-changelog']
+        self.ingest_request_source_whitelist = [
+            'fatcat-changelog',
+            'fatcat-ingest-container',
+        ]
         if kwargs.get('skip_source_whitelist', False):
             self.ingest_request_source_whitelist = []
 
     def want(self, row):
         """
-        Logic here probably needs work:
+        Logic here probably needs work (TODO):
 
         - Direct ingests via DOI from fatcat-changelog should probably go
           through regardless of GROBID status
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)