diff options
author | Bryan Newbold <bnewbold@robocracy.org> | 2019-11-13 16:41:25 -0800 |
---|---|---|
committer | Bryan Newbold <bnewbold@robocracy.org> | 2019-11-15 16:46:26 -0800 |
commit | 1d732653ed37bb4895b4702072591b776046c302 (patch) | |
tree | a9ebc70f738d5d4972b84ce3e32314cd23a68952 /python | |
parent | c59ab02bb9c9e486c98e758f0098be09c1973b42 (diff) | |
download | fatcat-1d732653ed37bb4895b4702072591b776046c302.tar.gz fatcat-1d732653ed37bb4895b4702072591b776046c302.zip |
ingest importer fixes
Diffstat (limited to 'python')
-rw-r--r-- | python/fatcat_tools/importers/ingest.py | 7 |
1 files changed, 4 insertions, 3 deletions
diff --git a/python/fatcat_tools/importers/ingest.py b/python/fatcat_tools/importers/ingest.py index 95df5efb..dbb8149e 100644 --- a/python/fatcat_tools/importers/ingest.py +++ b/python/fatcat_tools/importers/ingest.py @@ -28,7 +28,8 @@ class IngestFileResultImporter(EntityImporter): print("Requiring GROBID status == 200") else: print("NOT checking GROBID success") - self.project_whitelist = ['fatcat-changelog'] + #self.project_whitelist = ['fatcat-changelog'] + self.project_whitelist = [] def want(self, row): """ @@ -83,12 +84,12 @@ class IngestFileResultImporter(EntityImporter): continue release_ident = release.ident break - if not release and row.get('grobid'): + if not release_ident and row.get('grobid'): # try biblio-glutton extracted hit if row['grobid'].get('fatcat_ident'): release = row['grobid']['fatcat_ident'].split('_')[-1] - if not release: + if not release_ident: self.counts['skip-release-not-found'] += 1 return None |