diff options
| -rw-r--r-- | python/sandcrawler/ingest.py | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py index 6d8b162..35b37fc 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -76,8 +76,10 @@ class IngestFileWorker(SandcrawlerWorker): # temporary, until we implement specific fetch and 'petabox' output "://archive.org/", + "://www.archive.org/", "://web.archive.org/web/", "://openlibrary.org/", + "://www.openlibrary.org/", "://fatcat.wiki/", # Domain squats |