diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-01-21 11:34:37 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-01-21 11:34:39 -0800 |
| commit | 2e93f94c9ebba689dde252ca8f5b106765cece88 (patch) | |
| tree | db30238693a42249a57420c7fcc0b54d0d50f6d7 | |
| parent | 20291471b34ea559d2ea5d45f3b05884e54d179a (diff) | |
| download | sandcrawler-2e93f94c9ebba689dde252ca8f5b106765cece88.tar.gz sandcrawler-2e93f94c9ebba689dde252ca8f5b106765cece88.zip | |
re-enable figshare and zenodo crawling
For daily imports
| -rw-r--r-- | python/sandcrawler/ingest.py | 8 |
1 files changed, 0 insertions, 8 deletions
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/", |