diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-09-17 19:16:24 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-09-17 19:16:24 -0700 |
| commit | 351ab23bea9a7a95a9be2b18f65ef51564222a8c (patch) | |
| tree | 3bb2814dc72458c022c7bc25b72e78ece39a6baa /python/sandcrawler | |
| parent | d3f2b9f75da41d8dd2ff712e34a83472e259c3af (diff) | |
| download | sandcrawler-351ab23bea9a7a95a9be2b18f65ef51564222a8c.tar.gz sandcrawler-351ab23bea9a7a95a9be2b18f65ef51564222a8c.zip | |
ingest: small bugfix to print pdfextract status on SUCCESS
Diffstat (limited to 'python/sandcrawler')
| -rw-r--r-- | python/sandcrawler/ingest.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py index e8b3551..e8e517a 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -499,7 +499,7 @@ class IngestFileWorker(SandcrawlerWorker): print("[SUCCESS\t] sha1:{} grobid:{} pdfextract:{}".format( result.get('file_meta', {}).get('sha1hex'), result.get('grobid', {}).get('status_code'), - result.get('pdfextract', {}).get('status'), + result.get('pdf_meta', {}).get('status'), ), file=sys.stderr) return result |