class TestGatherMethods(HarvestFixtureBase):
def setup(self):
HarvestFixtureBase.setup(self)
# Create source
source_fixture = {
'url': u'http://127.0.0.1:8999/gemini2.1/dataset1.xml',
'type': u'gemini-single'
}
source, job = self._create_source_and_job(source_fixture)
self.harvester = GeminiHarvester()
self.harvester.harvest_job = job
def teardown(self):
model.repo.rebuild_db()
def test_get_gemini_string_and_guid(self):
res = self.harvester.get_gemini_string_and_guid(BASIC_GEMINI, url=None)
assert_equal(res, (BASIC_GEMINI, GUID))
def test_get_gemini_string_and_guid__no_guid(self):
res = self.harvester.get_gemini_string_and_guid(GEMINI_MISSING_GUID, url=None)
assert_equal(res, (GEMINI_MISSING_GUID, ''))
def test_get_gemini_string_and_guid__non_parsing(self):
content = '<gmd:MD_Metadata xmlns:gmd="http://www.isotc211.org/2005/gmd" xmlns:gco="http://www.isotc211.org/2005/gco">' # no closing tag
assert_raises(lxml.etree.XMLSyntaxError, self.harvester.get_gemini_string_and_guid, content)
def test_get_gemini_string_and_guid__empty(self):
content = ''
assert_raises(lxml.etree.XMLSyntaxError, self.harvester.get_gemini_string_and_guid, content)
def test_licence_url_multiple_urls(self):
# only the first URL is extracted
assert_equal(GeminiHarvester._extract_first_licence_url(
['Reference and PSMA Only',
'http://www.test.gov.uk/licenseurl',
'http://www.test.gov.uk/2nd_licenseurl']),
'http://www.test.gov.uk/licenseurl')
def test_responsible_organisation_basic(self):
responsible_organisation = [{'organisation-name': 'Ordnance Survey',
'role': 'owner'},
{'organisation-name': 'Maps Ltd',
'role': 'distributor'}]
assert_equal(GeminiHarvester._process_responsible_organisation(responsible_organisation),
('Ordnance Survey', ['Maps Ltd (distributor)',
'Ordnance Survey (owner)']))
def test_responsible_organisation_blank_provider(self):
# no owner or publisher, so blank provider
responsible_organisation = [{'organisation-name': 'Ordnance Survey',
'role': 'resourceProvider'},
{'organisation-name': 'Maps Ltd',
'role': 'distributor'}]
assert_equal(GeminiHarvester._process_responsible_organisation(responsible_organisation),
('', ['Maps Ltd (distributor)',
'Ordnance Survey (resourceProvider)']))
def test_responsible_organisation_publisher(self):
# no owner, so falls back to publisher
responsible_organisation = [{'organisation-name': 'Ordnance Survey',
'role': 'publisher'},
{'organisation-name': 'Maps Ltd',
'role': 'distributor'}]
assert_equal(GeminiHarvester._process_responsible_organisation(responsible_organisation),
('Ordnance Survey', ['Maps Ltd (distributor)',
'Ordnance Survey (publisher)']))
def setup(self):
HarvestFixtureBase.setup(self)
# Create source
source_fixture = {
'url': u'http://127.0.0.1:8999/gemini2.1/dataset1.xml',
'type': u'gemini-single'
}
source, job = self._create_source_and_job(source_fixture)
self.harvester = GeminiHarvester()
self.harvester.harvest_job = job
def test_responsible_organisation_multiple_roles(self):
# provider is the owner (ignores publisher)
responsible_organisation = [{'organisation-name': 'Ordnance Survey',
'role': 'publisher'},
{'organisation-name': 'Ordnance Survey',
'role': 'custodian'},
{'organisation-name': 'Distributor',
'role': 'distributor'}]
assert_equal(GeminiHarvester._process_responsible_organisation(responsible_organisation),
('Ordnance Survey', ['Distributor (distributor)',
'Ordnance Survey (publisher, custodian)',
]))
def test_responsible_organisation_owner(self):
# provider is the owner (ignores publisher)
responsible_organisation = [{'organisation-name': 'Ordnance Survey',
'role': 'publisher'},
{'organisation-name': 'Owner',
'role': 'owner'},
{'organisation-name': 'Maps Ltd',
'role': 'distributor'}]
assert_equal(GeminiHarvester._process_responsible_organisation(responsible_organisation),
('Owner', ['Owner (owner)',
'Maps Ltd (distributor)',
'Ordnance Survey (publisher)',
]))
def test_responsible_organisation_blank(self):
# no owner or publisher, so blank provider
responsible_organisation = []
assert_equal(GeminiHarvester._process_responsible_organisation(responsible_organisation),
('', []))
def test_licence_url_embedded_at_start(self):
# URL is embedded at the start of the text field and the
# whole field is returned. Noting this unusual behaviour
assert_equal(GeminiHarvester._extract_first_licence_url(
['http://www.test.gov.uk/licenseurl Reference and PSMA Only']),
'http://www.test.gov.uk/licenseurl Reference and PSMA Only')
def test_licence_url_embedded(self):
# URL is embedded within the text field and not extracted
assert_equal(GeminiHarvester._extract_first_licence_url(
['Reference and PSMA Only http://www.test.gov.uk/licenseurl']),
None)
def test_licence_url_normal(self):
assert_equal(GeminiHarvester._extract_first_licence_url(
['Reference and PSMA Only',
'http://www.test.gov.uk/licenseurl']),
'http://www.test.gov.uk/licenseurl')
