diff options
| author | Bryan Newbold <bnewbold@robocracy.org> | 2020-02-14 00:07:56 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@robocracy.org> | 2020-02-14 00:07:56 -0800 |
| commit | ed38bfde4e1eaddd6d710802b6f372c7b0aab26b (patch) | |
| tree | 7bf840c17fdc5b41e26239582b09255a6fc286f6 /python/tests/transform_elasticsearch.py | |
| parent | 2f8788152ff740d049d11e2e263cac978d526e2a (diff) | |
| download | fatcat-ed38bfde4e1eaddd6d710802b6f372c7b0aab26b.tar.gz fatcat-ed38bfde4e1eaddd6d710802b6f372c7b0aab26b.zip | |
ES updates: fix tests to accept archive.org in host/domain
Diffstat (limited to 'python/tests/transform_elasticsearch.py')
| -rw-r--r-- | python/tests/transform_elasticsearch.py | 5 |
1 files changed, 3 insertions, 2 deletions
diff --git a/python/tests/transform_elasticsearch.py b/python/tests/transform_elasticsearch.py index c94ab375..a954fc4d 100644 --- a/python/tests/transform_elasticsearch.py +++ b/python/tests/transform_elasticsearch.py @@ -110,9 +110,10 @@ def test_elasticsearch_file_transform(matched_importer): assert 'web' in es['rels'] assert 'www.zhros.ru' in es['hosts'] assert 'zhros.ru' in es['domains'] + assert 'archive.org' in (es['hosts'] + es['domains']) + assert 'web.archive.org' in (es['hosts'] + es['domains']) + # old regression assert not '.archive.org' in (es['hosts'] + es['domains']) - assert not 'archive.org' in (es['hosts'] + es['domains']) - assert not 'web.archive.org' in (es['hosts'] + es['domains']) def test_elasticsearch_changelog_transform(matched_importer): ce = entity_from_json(open('./tests/files/changelog_3469683.json', 'r').read(), ChangelogEntry) |