From 2e93f94c9ebba689dde252ca8f5b106765cece88 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Tue, 21 Jan 2020 11:34:37 -0800 Subject: re-enable figshare and zenodo crawling For daily imports --- python/sandcrawler/ingest.py | 8 -------- 1 file changed, 8 deletions(-) (limited to 'python') diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py index 94eb9a5..148abdb 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -65,14 +65,6 @@ class IngestFileWorker(SandcrawlerWorker): self.try_spn2 = kwargs.get('try_spn2', True) self.base_url_blocklist = [ - # temporary, until we do specific crawls - "://doi.org/10.5281/zenodo", - "://doi.org/10.6084/", - "://doi.org/10.11583/", - "://doi.org/10.1184/", - "://zenodo.org/", - "://figshare.com/", - # temporary, until we implement specific fetch and 'petabox' output "://archive.org/", "://web.archive.org/web/", -- cgit v1.2.3