def test_landing_page_is_missing_jsonld(self):
"""
SCENARIO: A landing page does not have a JSON-LD <SCRIPT> element.
EXPECTED RESULT: A SkipError is raised. Many CUAHSI documents
don't have JSON-LD, and we don't want that counting against the
failure count.
"""
package = 'tests.data.cuahsi.81e947faccf04de59392dddaac77bc75'
contents = ir.read_text(package, 'landing_page.no_json_ld.html')
doc = lxml.etree.HTML(contents)
obj = CUAHSIHarvester()
with self.assertLogs(logger=obj.logger, level='DEBUG'):
with self.assertRaises(JsonLdError):
obj.get_jsonld(doc)
def test_landing_page_jsonld(self):
"""
SCENARIO: A landing page does have a JSON-LD <SCRIPT> element.
EXPECTED RESULT: The JSON-LD is loaded.
"""
package = 'tests.data.cuahsi.81e947faccf04de59392dddaac77bc75'
contents = ir.read_text(package, 'landing_page.html')
doc = lxml.etree.HTML(contents)
obj = CUAHSIHarvester()
with self.assertLogs(logger=obj.logger, level='DEBUG'):
j = obj.get_jsonld(doc)
# To make sure we have JSON, feed it to something that expects it.
json.dumps(j)