diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-01-10 16:22:01 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-01-10 16:22:01 -0800 |
| commit | a54050c46654b0a6782abe5c0462c6f1c628ef2d (patch) | |
| tree | fcd81e12cfe22475bc267e49a9820a06f0e0cb3c | |
| parent | 566469783d6ea0e578d314232bff7d622c671241 (diff) | |
| download | sandcrawler-a54050c46654b0a6782abe5c0462c6f1c628ef2d.tar.gz sandcrawler-a54050c46654b0a6782abe5c0462c6f1c628ef2d.zip | |
limit length of error messages
| -rw-r--r-- | python/sandcrawler/ingest.py | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py index f618f1b..12d1473 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -178,19 +178,19 @@ class IngestFileWorker(SandcrawlerWorker): resource = self.find_resource(next_url, best_mimetype) except SavePageNowError as e: result['status'] = 'spn2-error' - result['error_message'] = str(e) + result['error_message'] = str(e)[:1600] return result except PetaboxError as e: result['status'] = 'petabox-error' - result['error_message'] = str(e) + result['error_message'] = str(e)[:1600] return result except CdxApiError as e: result['status'] = 'cdx-error' - result['error_message'] = str(e) + result['error_message'] = str(e)[:1600] return result except WaybackError as e: result['status'] = 'wayback-error' - result['error_message'] = str(e) + result['error_message'] = str(e)[:1600] return result if not resource.hit: |