def make_config(data_file):
return Config({
'output_dir': str(tmpdir),
'base_dir': os.path.join(base_dir, "."),
'data_file': os.path.join(base_dir, data_file),
'config_dir': os.path.join(base_dir, "config")
})
def make_config(data_file):
return Config({
'output_dir':
str(tmpdir),
'base_dir':
os.path.join(base_dir, "."),
'data_file':
os.path.join(base_dir, data_file),
'config_dir':
os.path.join(base_dir, "config"),
'ontology':
os.path.join(base_dir, 'config/ppo-reasoned-no-imports.owl')
})
def make_config(data_file):
ns = Namespace(chunk_size=50000,
config_dir='test/config',
data_file=os.path.join(base_dir, data_file),
drop_invalid=True,
input_dir='test/data',
log_file=False,
num_processes=4,
ontology='test/test-ontology.owl',
output_dir='test/data',
preprocessor=None,
project='test',
project_base='projects',
reasoner_config=None,
split_data_column=None,
verbose=True)
#Build the Config class
return Config(**ns.__dict__)
def make_config(data_file):
#ns = Namespace(chunk_size=50000, config_dir='test/config', data_file=os.path.join(base_dir, data_file), drop_invalid=True, input_dir='test/data/input', log_file=False, num_processes=4, ontology='https://raw.githubusercontent.com/PlantPhenoOntology/ppo/master/releases/2018-07-31/ppo.owl', output_dir='test/data/output', preprocessor=None, project='test', project_base='projects', reasoner_config=None, split_data_column=None, verbose=True)
ns = Namespace(chunk_size=50000,
config_dir='test/config',
data_file=os.path.join(base_dir, data_file),
drop_invalid=True,
input_dir='test/data/input',
log_file=False,
num_processes=4,
ontology='test/test-ontology.owl',
output_dir='test/data/output',
preprocessor=None,
project='test',
project_base='projects',
reasoner_config=None,
split_data_column=None,
verbose=True)
# Build the Config class
return Config(**ns.__dict__)
def test_config(tmpdir):
base_dir = os.path.dirname(__file__)
# Mimic the argument constructor from process:
ns = Namespace(chunk_size=50000,
config_dir='test/config',
data_file=None,
drop_invalid=True,
input_dir='test/data/input',
log_file=False,
num_processes=4,
ontology='test/test-ontology.owl',
output_dir='test/data/output',
preprocessor=None,
project='test',
project_base='projects',
reasoner_config=None,
split_data_column=None,
verbose=True)
# Build the Config class
config = Config(**ns.__dict__)
# Test that attributes are accessible
assert config.output_dir == 'test/data/output'
# verify that none existent attribute returns None
assert config.doesnt_exist is None
# verify rules were parsed 5 rules
assert len(config.rules) == 5
# should split | delimited rule columns
assert isinstance(config.rules[0]['columns'], list)
# should assign default error level
for rule in config.rules:
assert rule['level'] in ["error", "warning"]
def test_config(tmpdir):
base_dir = os.path.dirname(__file__)
config = Config(
{
'output_dir': str(tmpdir),
'base_dir': os.path.join(base_dir, "."),
'data_file': os.path.join(base_dir, "data/invalid_input.csv"),
'config_dir': os.path.join(base_dir, "config")
},
kw=False)
# verify that passed in args & kwargs are accessible as attributes
assert config.kw == False
assert config.output_dir == tmpdir
# should setup some attributes
# These two failing turning off for now
assert config.invalid_data_file == tmpdir.join('invalid_data.csv')
# verify that none existent attribute returns None
assert config.doesnt_exist is None
# verify rules were parsed 5 rules + 1 default
assert len(config.rules) == 6
# should split | delimited rule columns
assert isinstance(config.rules[0]['columns'], list)
# should assign default error level
for rule in config.rules:
assert rule['level'] in ["error", "warning"]
# should parse phenophase_descriptions file
descriptions = config.lists['phenophase_descriptions.csv']
assert {
'field': 'Reproductive',
'defined_by': 'http://purl.obolibrary.org/obo/PPO_0002025'
} in descriptions
assert {
'field': 'Flowering',
'defined_by': 'http://purl.obolibrary.org/obo/PPO_0002035'
} in descriptions
assert {
'field': 'Fruiting',
'defined_by': 'http://purl.obolibrary.org/obo/PPO_0002045'
} in descriptions
# should be 3 valid phenophase_descriptions list items
assert len(descriptions) == 3
assert len(config.lists['phenophase_descriptions.csv']) == 3
# should parse entities and perform label substitution
assert {
'alias':
'plantStructurePresence',
'concept_uri':
'http://www.w3.org/1999/02/22-rdf-syntax-ns#type',
'unique_key':
'record_id',
'identifier_root':
'http://n2t.net/ark:/21547/Anl2',
'columns': [('phenophase_name',
'http://www.w3.org/1999/02/22-rdf-syntax-ns#type')]
} in config.entities
assert {
'alias':
'phenologicalObservingProcess',
'concept_uri':
'http://purl.obolibrary.org/obo/BCO_0000003',
'unique_key':
'record_id',
'identifier_root':
'http://n2t.net/ark:/21547/Anm2',
'columns':
[('record_id', 'http://rs.tdwg.org/dwc/terms/EventID'),
('latitude', 'http://rs.tdwg.org/dwc/terms/decimalLatitude'),
('longitude', 'http://rs.tdwg.org/dwc/terms/decimalLongitude'),
('year', 'http://rs.tdwg.org/dwc/terms/year'),
('day_of_year', 'http://rs.tdwg.org/dwc/terms/startDayOfYear'),
('source', 'http://purl.org/dc/elements/1.1/creator')]
} in config.entities
assert len(config.entities) == 2
# should parse relations and perform label substitution
assert {
'subject_entity_alias': 'plantStructurePresence',
'predicate': 'http://purl.obolibrary.org/obo/OBI_0000295',
'object_entity_alias': 'phenologicalObservingProcess'
} in config.relations
def test_config(tmpdir):
base_dir = os.path.dirname(__file__)
config = Config(
{
'output_dir': str(tmpdir),
'base_dir': os.path.join(base_dir, "."),
'data_file': os.path.join(base_dir, "data/invalid_input.csv"),
'config_dir': os.path.join(base_dir, "config")
},
kw=False)
# verify that passed in args & kwargs are accessible as attributes
assert config.kw == False
assert config.output_dir == tmpdir
# should setup some attributes
# These two failing turning off for now
assert config.invalid_data_file == tmpdir.join('invalid_data.csv')
# verify that none existent attribute returns None
assert config.doesnt_exist is None
print(len(config.rules))
# verify rules were parsed 5 rules + 1 default
assert len(config.rules) == -1 # TODO: confirm rule length
# should split | delimited rule columns
assert isinstance(config.rules[0]['columns'], list)
# should assign default error level
for rule in config.rules:
assert rule['level'] in ["error", "warning"]
# should parse entities and perform label substitution
assert {
'alias': 'vertebrateOrganism',
'concept_uri': 'http://purl.obolibrary.org/obo/NCBITaxon_7742',
'unique_key': 'occurence_id',
'identifier_root': 'urn:vertOrganism',
'columns': [('genus', 'http://rs.tdwg.org/dwc/terms/genus')]
} in config.entities
assert {
'alias':
'organismalTraitObsProc',
'concept_uri':
'http://purl.obolibrary.org/obo/OVT_0000002',
'unique_key':
'occurence_id',
'identifier_root':
'urn:traitObsProc',
'columns':
[('record_id', 'http://rs.tdwg.org/dwc/terms/EventID'),
('latitude', 'http://rs.tdwg.org/dwc/terms/decimalLatitude'),
('longitude', 'http://rs.tdwg.org/dwc/terms/decimalLongitude'),
('year', 'http://rs.tdwg.org/dwc/terms/year'),
('event_date', 'http://rs.tdwg.org/dwc/terms/even')]
} in config.entities
assert len(config.entities) == 2
# should parse relations and perform label substitution
assert {
'subject_entity_alias': 'vertebrateOrganism',
'predicate': 'http://purl.obolibrary.org/obo/OBI_0000295',
'object_entity_alias': 'organismalTraitObsgProc'
} in config.relations