def test_iterate_order(self):
"""Test that the order of appearance in Core file is respected when iterating."""
# This is also probably tested indirectly elsewhere, but this is the right place :)
with DwCAReader(sample_data_path('dwca-ids.zip')) as dwca:
l = list(dwca)
# Row IDs are ordered like this in core file: id 4-1-3-2
self.assertEqual(int(l[0].id), 4)
self.assertEqual(int(l[1].id), 1)
self.assertEqual(int(l[2].id), 3)
self.assertEqual(int(l[3].id), 2)
def test_no_cr_left(self):
"""Test no carriage return characters are left at end of line."""
# We know we have no \n in our test archive, so if we fine one
# it's probably a character that was left by error when parsing
# line
with DwCAReader(sample_data_path('dwca-simple-test-archive.zip')) as simple_dwca:
for l in simple_dwca:
for k, v in l.data.items():
self.assertFalse(v.endswith("\n"))
def test_archives_without_metadata(self):
"""Ensure we can deal with an archive containing a metafile, but no metadata."""
with DwCAReader(sample_data_path('dwca-nometadata.zip')) as dwca:
self.assertIsNone(dwca.metadata)
# But the data is nevertheless accessible
rows = list(dwca)
self.assertEqual(len(rows), 2)
self.assertEqual('Borneo', rows[0].data[qn('locality')])
self.assertEqual('Mumbai', rows[1].data[qn('locality')])
def test_metadata(self):
"""A few basic tests on the metadata attribute."""
with DwCAReader(sample_data_path('dwca-simple-test-archive.zip')) as dwca:
# Assert metadata is an instance of ElementTree.Element
self.assertIsInstance(dwca.metadata, ET.Element)
# Assert we can read basic fields from EML:
v = (dwca.metadata.find('dataset').find('creator').find('individualName')
.find('givenName').text)
self.assertEqual(v, 'Nicolas')
def test_manual_cleanup_zipped(self):
"""Test no temporary files are left after execution (calling close() manually)."""
num_files_before = len(os.listdir('.'))
r = DwCAReader(sample_data_path('dwca-simple-test-archive.zip'))
r.close()
num_files_after = len(os.listdir('.'))
self.assertEqual(num_files_before, num_files_after)
def test_auto_cleanup_zipped(self):
"""Test no temporary files are left after execution (using 'with' statement)."""
num_files_before = len(os.listdir('.'))
with DwCAReader(sample_data_path('dwca-simple-test-archive.zip')):
pass
num_files_after = len(os.listdir('.'))
self.assertEqual(num_files_before, num_files_after)
def test_default_values_metafile(self):
"""
Ensure default values are used when optional attributes are absent in metafile.
Optional attributes tested here: linesTerminatedBy, fieldsTerminatedBy.
"""
with DwCAReader(sample_data_path('dwca-meta-default-values')) as dwca:
# Test iterating on rows...
for row in dwca:
self.assertIsInstance(row, CoreRow)
def test_source_is_dwca(self):
print 'testing source_is_dwca'
dwca = self.framework.dwca
dwcareader = None
try:
dwcareader = DwCAReader(dwca)
except:
dwcareader = None
s = 'No viable Darwin Core archive found at %s' % dwca
self.assertIsNotNone(dwcareader, s)
def test_exposes_core_type(self):
"""Test that it exposes the Archive Core Type as type"""
with DwCAReader(BASIC_ARCHIVE_PATH) as dwca:
coredescriptor = dwca.descriptor.core
# dwca-simple-test-archive.zip should be of Occurrence type
self.assertEqual(coredescriptor.type,
'http://rs.tdwg.org/dwc/terms/Occurrence')
# Check that shortcuts also work
self.assertEqual(coredescriptor.type, qn('Occurrence'))
def test_implicit_encoding_metadata(self):
"""If the metadata file doesn't specifies encoding, use UTF-8."""
with DwCAReader(sample_data_path("dwca-simple-dir")) as dwca:
v = (
dwca.metadata.find("dataset")
.find("creator")
.find("individualName")
.find("surName")
.text
)
self.assertEqual(v, u"Noé")
