class TestShulmanFinder(unittest.TestCase):
def setUp(self):
self.daoist = Daoist('shulman',
'test/fixtures/Shulman_SeriesII_handles.csv',
'test/fixtures/Shulman_SeriesI-III.xml', 'test_ead.xml', 2)
self.daoist.load_handles()
self.daoist.load_ead()
self.find = finders.shulman
self.handle = {
'title': 'Job 2892: Solomon R. Guggenheim Museum (New York, N.Y.), 1959',
'handle': 'http://hdl.handle.net/10020/2004r10_job2892'
}
self.bad_handle = {
'title': 'not a real handle',
'handle': 'http://hdl.handle.net/10020/2014r20e00xx'
}
self.test_node = self.daoist.src_root.xpath('//ead:c03[@id="ref6695"]',
namespaces=self.daoist.src_nsmap)[0]
def test_finder(self):
hits = self.find(self.daoist.src_root, self.daoist.src_nsmap,
self.handle, 2)
self.assertEqual(len(hits), 1)
self.assertEqual(self.test_node.text, hits[0].text)
hits = self.find(self.daoist.src_root, self.daoist.src_nsmap,
self.bad_handle, 2)
self.assertEqual(len(hits), 0)
class TestShunkFinder(unittest.TestCase):
def setUp(self):
self.daoist = Daoist('shunk', 'test/fixtures/Shunk_handles.csv',
'test/fixtures/Shunk_ead.xml', 'test_ead.xml')
self.daoist.load_handles()
self.daoist.load_ead()
self.find = finders.shunk
self.handle = {
'title': 'Yves Klein, first experiments with "Living Brushes," ' +
'Robert Godet\'s apartment, 9 rue Le-Regrattier, Île ' +
'Saint-Louis, Paris',
'handle': 'http://hdl.handle.net/10020/2014r20e0012'
}
self.bad_handle = {
'title': 'not a real handle',
'handle': 'http://hdl.handle.net/10020/2014r20e00xx'
}
self.test_node = self.daoist.src_root.xpath('//ead:c02[@id="ref18"]',
namespaces=self.daoist.src_nsmap)[0]
def test_finder(self):
hits = self.find(self.daoist.src_root, self.daoist.src_nsmap,
self.handle)
self.assertEqual(len(hits), 1)
hits = self.find(self.daoist.src_root, self.daoist.src_nsmap,
self.bad_handle)
self.assertEqual(len(hits), 0)
def setUp(self):
self.daoist = Daoist('shunk', 'test/fixtures/Shunk_handles_FAKE.csv',
'test/fixtures/Shunk_ead_FAKE.xml', 'test_dir')
self.daoist.load_handles()
self.daoist.load_ead()
self.test_node = self.daoist.src_root.xpath('//ead:c02[@id="ref18"]',
namespaces=self.daoist.src_nsmap)[0]
self.handle = {
'title': 'Yves Klein, first experiments with "Living Brushes," ' +
'Robert Godet\'s apartment, 9 rue Le-Regrattier, Île ' +
'Saint-Louis, Paris',
'handle': 'http://hdl.handle.net/10020/2014r20e0012'
}
def setUp(self):
self.daoist = Daoist('clark_duveen', 'test/fixtures/ClarkDuveen_handles.csv',
'test/fixtures/clark_duveen_ead.xml', 'test_ead.xml', by_id=True)
self.daoist.load_handles()
self.daoist.load_ead()
self.find = finders.by_id
self.handle = {
'handle': 'http://hdl.handle.net/10020/2007d1_d0045'
}
self.bad_handle = {
'handle': 'http://hdl.handle.net/10020/2007d1_d00xx'
}
self.test_node = self.daoist.src_root.xpath('//ead:c03[@id="aspace_ref162_gsw"]',
namespaces=self.daoist.src_nsmap)[0]
class TestDuveenFinder(unittest.TestCase):
def setUp(self):
self.daoist = Daoist('duveen', 'test/fixtures/duveen.csv',
'test/fixtures/duveen_ead.xml', 'test_ead.xml')
self.daoist.load_handles()
self.daoist.load_ead()
self.find = finders.duveen
self.handle = {
'title': 'TV film,',
'handle': 'http://hdl.handle.net/10020/960015b413f008'
}
self.bad_handle = {
'title': 'not a real handle',
'handle': 'http://hdl.handle.net/10020/960015b999f999'
}
self.test_node = self.daoist.src_root.xpath('//ead:c05[@id="ref19737"]',
namespaces=self.daoist.src_nsmap)[0]
def test_finder(self):
hits = self.find(self.daoist.src_root, self.daoist.src_nsmap,
self.handle)
self.assertEqual(len(hits), 1)
hits = self.find(self.daoist.src_root, self.daoist.src_nsmap,
self.bad_handle)
self.assertEqual(len(hits), 0)
def test_ranged_values(self):
handle = {
'title': 'Invoices, June 1, 1936-May 31, 1937, cont.',
'handle': 'http://hdl.handle.net/10020/960015b139f006'
}
hits = self.find(self.daoist.src_root, self.daoist.src_nsmap, handle)
self.assertEqual(len(hits), 1)
test_node = self.daoist.src_root.xpath('ead:archdesc/ead:dsc/ead:c01/ead:c02/ead:c03/ead:c04[@id="ref592"]',
namespaces=self.daoist.src_nsmap)[0]
self.assertEqual(test_node.text, hits[0].text)
class TestDaoistInit(unittest.TestCase):
def setUp(self):
self.daoist = Daoist('shunk', 'test/fixtures/Shunk_handles.csv',
'test/fixtures/Shunk_ead.xml', 'test_ead')
def test_load_handles(self):
self.daoist.load_handles()
handle_list = self.daoist.handles
self.assertEqual(len(handle_list), 1069)
self.assertEqual(handle_list[0]['title'],
'Harry Shunk street photography, "tuileries," Paris')
self.assertEqual(handle_list[0]['handle'],
'http://hdl.handle.net/10020/2014r20e0001')
def test_load_ead(self):
self.daoist.load_ead()
self.assertEqual(self.daoist.src_root.tag,
'{urn:isbn:1-931666-22-9}ead')
def test_series_error(self):
with self.assertRaises(SeriesError):
self.daoist = Daoist('shulman', 'test/fixtures/Shulman_handles.csv',
'test/fixtures/Shulman_SeriesI-III.xml', 'test_ead')
class TestByIdFinder(unittest.TestCase):
def setUp(self):
self.daoist = Daoist('clark_duveen', 'test/fixtures/ClarkDuveen_handles.csv',
'test/fixtures/clark_duveen_ead.xml', 'test_ead.xml', by_id=True)
self.daoist.load_handles()
self.daoist.load_ead()
self.find = finders.by_id
self.handle = {
'handle': 'http://hdl.handle.net/10020/2007d1_d0045'
}
self.bad_handle = {
'handle': 'http://hdl.handle.net/10020/2007d1_d00xx'
}
self.test_node = self.daoist.src_root.xpath('//ead:c03[@id="aspace_ref162_gsw"]',
namespaces=self.daoist.src_nsmap)[0]
def test_finder(self):
hits = self.find(self.daoist.src_root, self.daoist.src_nsmap,
self.handle)
self.assertEqual(len(hits), 1)
hits = self.find(self.daoist.src_root, self.daoist.src_nsmap,
self.bad_handle)
self.assertEqual(len(hits), 0)
def setUp(self):
self.daoist = Daoist('duveen', 'test/fixtures/duveen.csv',
'test/fixtures/duveen_ead.xml', 'test_ead.xml')
self.daoist.load_handles()
self.daoist.load_ead()
self.find = finders.duveen
self.handle = {
'title': 'TV film,',
'handle': 'http://hdl.handle.net/10020/960015b413f008'
}
self.bad_handle = {
'title': 'not a real handle',
'handle': 'http://hdl.handle.net/10020/960015b999f999'
}
self.test_node = self.daoist.src_root.xpath('//ead:c05[@id="ref19737"]',
namespaces=self.daoist.src_nsmap)[0]
def setUp(self):
self.daoist = Daoist('shulman',
'test/fixtures/Shulman_SeriesII_handles.csv',
'test/fixtures/Shulman_SeriesI-III.xml', 'test_ead.xml', 2)
self.daoist.load_handles()
self.daoist.load_ead()
self.find = finders.shulman
self.handle = {
'title': 'Job 2892: Solomon R. Guggenheim Museum (New York, N.Y.), 1959',
'handle': 'http://hdl.handle.net/10020/2004r10_job2892'
}
self.bad_handle = {
'title': 'not a real handle',
'handle': 'http://hdl.handle.net/10020/2014r20e00xx'
}
self.test_node = self.daoist.src_root.xpath('//ead:c03[@id="ref6695"]',
namespaces=self.daoist.src_nsmap)[0]
class TestDaoistProcessing(unittest.TestCase):
def setUp(self):
self.daoist = Daoist('shunk', 'test/fixtures/Shunk_handles_FAKE.csv',
'test/fixtures/Shunk_ead_FAKE.xml', 'test_dir')
self.daoist.load_handles()
self.daoist.load_ead()
self.test_node = self.daoist.src_root.xpath('//ead:c02[@id="ref18"]',
namespaces=self.daoist.src_nsmap)[0]
self.handle = {
'title': 'Yves Klein, first experiments with "Living Brushes," ' +
'Robert Godet\'s apartment, 9 rue Le-Regrattier, Île ' +
'Saint-Louis, Paris',
'handle': 'http://hdl.handle.net/10020/2014r20e0012'
}
def test_check_for_dupe(self):
namespaces = self.daoist.src_nsmap
self.duped_testnode = self.daoist.src_root.xpath('//ead:c02[@id="ref12"]', namespaces =namespaces)[0]
handle = {'title': 'Allan Kaprow artwork, unidentified', 'handle': 'http://hdl.handle.net/10020/2014r20e0002'}
dupecheck = self.daoist.check_for_dupe(self.duped_testnode, namespaces, handle)
self.assertEqual(dupecheck, True)
dupecheck = self.daoist.check_for_dupe(self.test_node, namespaces, self.handle)
self.assertEqual(dupecheck, False)
def test_get_title(self):
title = self.daoist.get_title(self.test_node)
self.assertEqual(title,
"Yves Klein, first experiments with Living Brushes, Robert " +
"Godet's apartment, 9 rue Le-Regrattier, Île Saint-Louis, Paris")
def test_add_dao(self):
self.daoist.add_dao(self.test_node, self.handle)
children = self.test_node.getchildren()
self.assertEqual(len(children), 4)
dao_hits = self.test_node.xpath(
'ead:dao[@ns2:href="http://hdl.handle.net/10020/2014r20e0012"]',
namespaces=self.daoist.src_nsmap)
self.assertEqual(len(dao_hits), 1)
def test_find_position(self):
pos = self.daoist.find_position(self.test_node)
self.assertEqual(pos, 1)
self.test_node2 = self.daoist.src_root.xpath('//ead:c02[@id="ref20"]',
namespaces=self.daoist.src_nsmap)[0]
pos = self.daoist.find_position(self.test_node2)
self.assertEqual(pos, 2)
self.test_node3 = self.daoist.src_root.xpath('//ead:c02[@id="ref24"]',
namespaces=self.daoist.src_nsmap)[0]
pos = self.daoist.find_position(self.test_node3)
self.assertEqual(pos, 2)
def test_process(self):
totals = self.daoist.process()
self.assertTrue(os.path.exists('test_dir'))
ts = self.daoist.timestamp
self.assertTrue(os.path.exists('test_dir/shunk_%s_ead_dao.xml' % ts))
self.assertEqual(totals, [1064, 2, 1, 4]) #I added 4 duplicates.
shutil.rmtree(self.daoist.target_dir, ignore_errors=True)
self.assertFalse(os.path.exists(self.daoist.target_dir))
def test_series_error(self):
with self.assertRaises(SeriesError):
self.daoist = Daoist('shulman', 'test/fixtures/Shulman_handles.csv',
'test/fixtures/Shulman_SeriesI-III.xml', 'test_ead')
def setUp(self):
self.daoist = Daoist('shunk', 'test/fixtures/Shunk_handles.csv',
'test/fixtures/Shunk_ead.xml', 'test_ead')
