diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-05-19 11:36:58 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-05-19 11:36:58 -0700 |
| commit | 3ee18580dd108c69c01cdf838a7f1a7d3d181629 (patch) | |
| tree | 3926506bea54c596e3df61c3c047f2c14ac2c5c1 /fatcat_scholar/work_pipeline.py | |
| parent | c5faf30df8772af607689a3f65241f34b365325b (diff) | |
| download | fatcat-scholar-3ee18580dd108c69c01cdf838a7f1a7d3d181629.tar.gz fatcat-scholar-3ee18580dd108c69c01cdf838a7f1a7d3d181629.zip | |
WIP on SIM pipeline
Diffstat (limited to 'fatcat_scholar/work_pipeline.py')
| -rw-r--r-- | fatcat_scholar/work_pipeline.py | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/fatcat_scholar/work_pipeline.py b/fatcat_scholar/work_pipeline.py index 161327a..b9dcbe8 100644 --- a/fatcat_scholar/work_pipeline.py +++ b/fatcat_scholar/work_pipeline.py @@ -98,7 +98,7 @@ class WorkPipeline(): if not fe.urls: return None grobid_meta = self.sandcrawler_db_client.get_grobid(fe.sha1) - if grobid_meta['status'] != 'success': + if not grobid_meta or grobid_meta['status'] != 'success': return None #print(grobid_meta) grobid_xml = self.sandcrawler_s3_client.get_blob( @@ -290,7 +290,7 @@ def main(): """ Run this command like: - python -m fatcat_scholar.release_pipeline + python -m fatcat_scholar.work_pipeline """ parser = argparse.ArgumentParser( |