diff options
Diffstat (limited to 'python/sandcrawler/ingest.py')
-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 c9a697c..4159e26 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -354,6 +354,7 @@ class IngestFileWorker(SandcrawlerWorker): return result next_url = fulltext_url.get('pdf_url') or fulltext_url.get('next_url') assert next_url + next_url = clean_url(next_url) print("[PARSE\t] {}\t{}".format( fulltext_url.get('technique'), next_url, |