def test_value_error_raised_when_metadata_not_present(self):
'''
Metadata is not present an exception is raised.
'''
d = Dataset(self.dataset_error).info()
with self.assertRaises(ValueError):
parse_dataset(d)
def test_metadata_is_complete(self):
'''
Metadata property is complete.
'''
d = Dataset(self.dataset_id).info()
result = parse_dataset(d)
for key in self.metadata_types.keys():
print(result['metadata'])
self.assertIs(type(result['metadata'][key]), self.metadata_types[key])
def test_parser_returns_metadata_and_resource(self):
'''
Dataset and the resource objects are returned.
'''
d = Dataset(self.dataset_id).info()
result = parse_dataset(d)
self.assertIs(type(result), type({}))
for key in ['metadata', 'resources']:
self.assertIn(key, result.keys())
def main():
'''
Program wrapper.
'''
print('%s Creating Dataverse instance.' % item('bullet'))
d = Dataverse('dataverse.harvard.edu', 'IFPRI')
print('%s Collecting all content from Dataverse.' % item('bullet'))
contents = d.contents()
#
# Collects data and organizes
# in lists and dictionaries.
#
datasets = []
resources = []
for dataset in contents:
print('%s Collecting data for ID %s.' % (item('bullet'), dataset['id']))
o = Dataset(dataset['id']).info()
if o.get('status') is 'ERROR' or None:
continue
try:
parsed_data = parse_dataset(o)
except ValueError:
print('%s Missing metadata. Not parsing.' % item('warn'))
continue
else:
datasets.append(parsed_data['metadata'])
resources += parsed_data['resources']
#
# Exports JSON files to disk..
#
export_json(datasets, 'data/datasets.json')
export_json(resources, 'data/resources.json')
print('%s Total datasets downloaded %s' % (item('success'), str(len(datasets))))
