diff options
author | Martin Czygan <martin.czygan@gmail.com> | 2019-12-27 00:24:35 +0100 |
---|---|---|
committer | Martin Czygan <martin.czygan@gmail.com> | 2019-12-27 00:24:35 +0100 |
commit | 8b447f75a38e595248bd065f74dc10a9ce918e99 (patch) | |
tree | a8ffa12f01cc4140d9b111d397f1dba9efb6ccee /python/tests/harvest_datacite.py | |
parent | 0f86388578fd7b0475f4e621f13f65b600ae2655 (diff) | |
download | fatcat-8b447f75a38e595248bd065f74dc10a9ce918e99.tar.gz fatcat-8b447f75a38e595248bd065f74dc10a9ce918e99.zip |
datacite: add simple test and fixture for datacite api interaction
Diffstat (limited to 'python/tests/harvest_datacite.py')
-rw-r--r-- | python/tests/harvest_datacite.py | 45 |
1 files changed, 45 insertions, 0 deletions
diff --git a/python/tests/harvest_datacite.py b/python/tests/harvest_datacite.py new file mode 100644 index 00000000..5aaa8ac4 --- /dev/null +++ b/python/tests/harvest_datacite.py @@ -0,0 +1,45 @@ + +import json +import pytest +import datetime +import responses +from fatcat_tools.harvest import * + + +@responses.activate +def test_datacite_harvest_date(mocker): + + # mock out the harvest state object so it doesn't try to actually connect + # to Kafka + mocker.patch('fatcat_tools.harvest.harvest_common.HarvestState.initialize_from_kafka') + + # mock day request to crossref API + with open('tests/files/datacite_api.json', 'r') as f: + crossref_resp = json.loads(f.readline()) + responses.add(responses.GET, 'https://api.datacite.org/dois', + json=crossref_resp, status=200) + + harvester = HarvestDataciteWorker( + kafka_hosts="dummy", + produce_topic="dummy-produce-topic", + state_topic="dummy-state-topic", + contact_email="test@fatcat.wiki", + ) + + harvester.producer = mocker.Mock() + + harvester.fetch_date(datetime.date(2019, 2, 3)) + + assert len(responses.calls) == 1 + + # ensure email was included in User-Agent + assert "mailto:test@fatcat.wiki" in responses.calls[0].request.headers['User-Agent'] + + # check that correct date param was passed as expected + assert "query=updated%3A%5B2019-02-03T00%3A00%3A00.000Z+TO+2019-02-03T23%3A59%3A59.999Z%5D" in responses.calls[0].request.url + + # check that we published the expected number of DOI objects were published + # to the (mock) kafka topic + assert harvester.producer.produce.call_count == 25 + assert harvester.producer.flush.call_count == 1 + assert harvester.producer.poll.called_once_with(0) |