From 36bfdda186151972167c35cd0f642bc0c2ca1a54 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Thu, 30 May 2019 13:05:22 -0700 Subject: fix and workaround container entities in release topic --- python/fatcat_tools/workers/changelog.py | 4 ++-- python/fatcat_tools/workers/elasticsearch.py | 6 ++++++ 2 files changed, 8 insertions(+), 2 deletions(-) (limited to 'python/fatcat_tools') diff --git a/python/fatcat_tools/workers/changelog.py b/python/fatcat_tools/workers/changelog.py index 5ef1c69c..a753bb49 100644 --- a/python/fatcat_tools/workers/changelog.py +++ b/python/fatcat_tools/workers/changelog.py @@ -75,8 +75,8 @@ class EntityUpdatesWorker(FatcatWorker): def run(self): changelog_topic = self.kafka.topics[self.consume_topic] release_topic = self.kafka.topics[self.release_topic] - file_topic = self.kafka.topics[self.release_topic] - container_topic = self.kafka.topics[self.release_topic] + file_topic = self.kafka.topics[self.file_topic] + container_topic = self.kafka.topics[self.container_topic] consumer = changelog_topic.get_balanced_consumer( consumer_group=self.consumer_group, diff --git a/python/fatcat_tools/workers/elasticsearch.py b/python/fatcat_tools/workers/elasticsearch.py index fc96e3d4..a18ed244 100644 --- a/python/fatcat_tools/workers/elasticsearch.py +++ b/python/fatcat_tools/workers/elasticsearch.py @@ -43,6 +43,12 @@ class ElasticsearchReleaseWorker(FatcatWorker): for msg in consumer: json_str = msg.value.decode('utf-8') + # HACK: work around a bug where container entities got published to + # release_v03 topic + if self.elasticsearch_document_name == "release": + entity_dict = json.loads(json_str) + if entity_dict.get('name') and not entity_dict.get('title'): + continue entity = entity_from_json(json_str, self.entity_type, api_client=ac) #print(entity) elasticsearch_endpoint = "{}/{}/{}/{}".format( -- cgit v1.2.3