diff options
| -rw-r--r-- | python/sandcrawler/ingest.py | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py index 4732509..38ba55c 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -223,6 +223,7 @@ class IngestFileWorker(SandcrawlerWorker): for block in self.base_url_blocklist: if block in base_url: + print("[SKIP {}\t] {}".format(ingest_type, base_url), file=sys.stderr) return dict(request=request, hit=False, status="skip-url-blocklist") print("[INGEST {}\t] {}".format(ingest_type, base_url), file=sys.stderr) |