def test_landing_page_is_not_published(self):
"""
SCENARIO: A landing page indicates that it is not published.
EXPECTED RESULT: A SkipError is raised.
"""
package = 'tests.data.cuahsi.81e947faccf04de59392dddaac77bc75'
contents = ir.read_text(package, 'landing_page.not_published.html')
doc = lxml.etree.HTML(contents)
obj = CUAHSIHarvester()
with self.assertLogs(logger=obj.logger, level='DEBUG'):
with self.assertRaises(SkipError):
obj.preprocess_landing_page(doc)
def test_landing_page_does_not_have_sharing_status_element(self):
"""
SCENARIO: A landing page does not have a sharing status element.
Such an element is needed because we only wish to harvest PUBLISHED
documents.
EXPECTED RESULT: A SkipError is raised.
"""
package = 'tests.data.cuahsi.81e947faccf04de59392dddaac77bc75'
contents = ir.read_text(package, 'landing_page.no_sharing_status.html')
doc = lxml.etree.HTML(contents)
obj = CUAHSIHarvester()
with self.assertLogs(logger=obj.logger, level='DEBUG'):
with self.assertRaises(SkipError):
obj.preprocess_landing_page(doc)