diff options
author | Bryan Newbold <bnewbold@archive.org> | 2020-01-14 16:05:41 -0800 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2020-01-14 16:05:41 -0800 |
commit | ba6f16a02cfde0e4acb499c00b456b42472c0b00 (patch) | |
tree | 9b27d9982e61ee97a82181249c0f9723980a4beb /python/tests/test_grobid.py | |
parent | 21599839802b8ef3a84ffe90855f7bceaaa12a0d (diff) | |
download | sandcrawler-ba6f16a02cfde0e4acb499c00b456b42472c0b00.tar.gz sandcrawler-ba6f16a02cfde0e4acb499c00b456b42472c0b00.zip |
more ftp status 226 support
Diffstat (limited to 'python/tests/test_grobid.py')
-rw-r--r-- | python/tests/test_grobid.py | 8 |
1 files changed, 6 insertions, 2 deletions
diff --git a/python/tests/test_grobid.py b/python/tests/test_grobid.py index 8c5e080..24ea40e 100644 --- a/python/tests/test_grobid.py +++ b/python/tests/test_grobid.py @@ -64,8 +64,12 @@ def test_grobid_worker_cdx(grobid_client, wayback_client): body=REAL_TEI_XML, content_type='text/xml') with open('tests/files/example.cdx', 'r') as cdx_file: - pusher = CdxLinePusher(worker, cdx_file, - filter_http_statuses=[200], filter_mimetypes=['application/pdf']) + pusher = CdxLinePusher( + worker, + cdx_file, + filter_http_statuses=[200, 226], + filter_mimetypes=['application/pdf'], + ) pusher_counts = pusher.run() assert pusher_counts['total'] assert pusher_counts['pushed'] == 7 |