diff options
author | Bryan Newbold <bnewbold@robocracy.org> | 2020-02-26 11:38:51 -0800 |
---|---|---|
committer | Bryan Newbold <bnewbold@robocracy.org> | 2020-02-26 11:38:51 -0800 |
commit | c57a743cb8b774750c99c6f079438666a87f6476 (patch) | |
tree | 0adace2adf58c40b4baa977e02bbf204e109842b /python | |
parent | 4e6bc246d01183f4c7ffad7d0d474e683f04c07f (diff) | |
download | fatcat-c57a743cb8b774750c99c6f079438666a87f6476.tar.gz fatcat-c57a743cb8b774750c99c6f079438666a87f6476.zip |
bulk ES transform: skip non-active entities
Diffstat (limited to 'python')
-rwxr-xr-x | python/fatcat_transform.py | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/python/fatcat_transform.py b/python/fatcat_transform.py index 42d2ea99..9ddbaa4d 100755 --- a/python/fatcat_transform.py +++ b/python/fatcat_transform.py @@ -30,6 +30,8 @@ def run_elasticsearch_releases(args): if not line: continue entity = entity_from_json(line, ReleaseEntity, api_client=args.api.api_client) + if entity['state'] != 'active': + continue args.json_output.write( json.dumps(release_to_elasticsearch(entity)) + '\n') @@ -39,6 +41,8 @@ def run_elasticsearch_containers(args): if not line: continue entity = entity_from_json(line, ContainerEntity, api_client=args.api.api_client) + if entity['state'] != 'active': + continue args.json_output.write( json.dumps(container_to_elasticsearch(entity)) + '\n') @@ -48,6 +52,8 @@ def run_elasticsearch_files(args): if not line: continue entity = entity_from_json(line, FileEntity, api_client=args.api.api_client) + if entity['state'] != 'active': + continue args.json_output.write( json.dumps(file_to_elasticsearch(entity)) + '\n') @@ -66,6 +72,8 @@ def run_citeproc_releases(args): if not line: continue entity = entity_from_json(line, ReleaseEntity, api_client=args.api.api_client) + if entity['state'] != 'active': + continue csl_json = release_to_csl(entity) csl_json['id'] = "release:" + (entity.ident or "unknown") out = citeproc_csl(csl_json, args.style, args.html) |