def test_next():
tsv_reader = TSVReader('data/example_samples.tsv',
'data/T2D_xls2xml_v1.conf', 'Sample')
row = tsv_reader.next()
assert isinstance(row, dict)
assert 0 == cmp(
row, {
'Novel Attributes': None,
'Ethnicity Description': None,
'Description': 'Male normal',
'Cell Type': 'Blood',
'Maternal_id': 'SAM111113',
'Center_name': 'WTGC cambridge',
'Gender': 'male',
'Subject_ID': 'SAM111111',
'Paternal_id': 'SAM111115',
'T2D': 0,
'Hispanic or Latino; of Spanish origin': None,
'Cohort ID': 'CO1111',
'Year of Birth': '1986',
'Age': '31',
'Analysis_alias': 'AN001',
'Sample_ID': 'SAM111111',
'Geno_ID': None,
'Year of first visit': None,
'Case_Control': 'Control',
'Ethnicity': 'EUWH',
'Phenotype': 'MeSH:D006262'
})
for row in tsv_reader:
assert isinstance(row, dict)
def test_get_current_headers():
tsv_reader = TSVReader('data/example_samples.tsv',
'data/T2D_xls2xml_v1.conf', 'Sample')
headers = tsv_reader.get_current_headers()
assert isinstance(headers, list)
assert set(headers) == {
'Sample_ID', 'Subject_ID', 'Geno_ID', 'Phenotype', 'Gender',
'Analysis_alias', 'Cohort ID', 'Ethnicity', 'Ethnicity Description',
'T2D', 'Case_Control', 'Description', 'Center_name',
'Hispanic or Latino; of Spanish origin', 'Age', 'Year of Birth',
'Year of first visit', 'Cell Type', 'Maternal_id', 'Paternal_id',
'Novel Attributes'
}
def test_validate_file():
validation_schema = 'data/T2D_xls2xml_v1.schema'
tsv_reader = TSVReader('data/example_samples.tsv', 'data/T2D_xls2xml_v1.conf', 'Sample')
assert utils.validate_file(tsv_reader, validation_schema)
xls_reader = XLSReader('data/example_AMP_T2D_Submission_form_V2.xlsx', 'data/T2D_xls2xml_v1.conf')
xls_reader.set_current_conf_key('Sample')
assert utils.validate_file(xls_reader, validation_schema)
def test_multiple_tsvs_to_xml():
analysis_tsv_reader = TSVReader('data/example_analysis.tsv', 'data/T2D_xls2xml_v1.conf', 'Analysis')
files_tsv_reader = TSVReader('data/example_files.tsv', 'data/T2D_xls2xml_v1.conf', 'File')
tsv_readers = [ ('Analysis', analysis_tsv_reader), ('File', files_tsv_reader) ]
output_xml = utils.multiple_objects_to_xml(tsv_readers, 'data/T2D_xls2xml_v1.schema',
'data/T2D_xls2xml_v2.xslt')
with open('data/example_analysis.xml', 'r') as analysis_example:
assert analysis_example.readline()
assert etree.tostring(output_xml, pretty_print=True) == analysis_example.read()
with pytest.raises(Exception):
tsv_readers = [
('Exception', TSVReader('data/example_analysis.tsv', 'data/T2D_xls2xml_v1.conf', 'Exception')),
('Expected', TSVReader('data/example_analysis.tsv', 'data/T2D_xls2xml_v1.conf', 'Expected'))
]
utils.multiple_objects_to_xml(tsv_readers, 'data/T2D_xls2xml_v1.schema',
'data/T2D_xls2xml_v2.xslt')
def test_get_valid_conf_keys():
tsv_reader = TSVReader('data/example_samples.tsv',
'data/T2D_xls2xml_v1.conf', 'Sample')
assert set(tsv_reader.get_valid_conf_keys()) == {'Sample'}
tsv_reader = TSVReader('data/example_samples.tsv',
'data/T2D_xls2xml_v1.conf', 'Analysis')
assert tsv_reader.get_valid_conf_keys() == []
def test_valid_data():
validator = MetadataValidator('data/T2D_xls2xml_v1.schema')
reader = XLSReader('data/example_AMP_T2D_Submission_form_V2.xlsx',
'data/T2D_xls2xml_v1.conf')
reader.active_worksheet = 'Sample'
row = reader.next()
assert validator.validate_data(row, 'Sample')
reader.active_worksheet = 'Analysis'
row = reader.next()
assert validator.validate_data(row, 'Analysis')
reader = TSVReader('data/example_samples.tsv', 'data/T2D_xls2xml_v1.conf',
'Sample')
row = reader.next()
assert validator.validate_data(row, 'Sample')
def test_extract_rows():
validation_schema = 'data/T2D_xls2xml_v1.schema'
rows = []
tsv_reader = TSVReader('data/example_samples.tsv', 'data/T2D_xls2xml_v1.conf', 'Sample')
assert utils.extract_rows(tsv_reader, 'Sample', validation_schema, rows)
assert isinstance(rows, list)
assert 6 == len(rows)
tsv_reader = TSVReader('data/example_samples.tsv', 'data/T2D_xls2xml_v1.conf', 'Sample')
for a, b in zip(rows, tsv_reader):
assert 0 == cmp(a, b)
rows = []
xls_reader = XLSReader('data/example_AMP_T2D_Submission_form_V2.xlsx', 'data/T2D_xls2xml_v1.conf')
assert utils.extract_rows(xls_reader, 'Sample', validation_schema, rows)
assert isinstance(rows, list)
assert 6 == len(rows)
xls_reader = XLSReader('data/example_AMP_T2D_Submission_form_V2.xlsx', 'data/T2D_xls2xml_v1.conf')
xls_reader.set_current_conf_key('Sample')
for a, b in zip(rows, xls_reader):
assert 0 == cmp(a, b)
rows = []
assert not utils.extract_rows(xls_reader, 'FalseExpected', validation_schema, rows)
def test_is_valid():
tsv_reader = TSVReader('data/example_samples.tsv',
'data/T2D_xls2xml_v1.conf', 'Sample')
assert tsv_reader.is_valid()
def test_is_not_valid():
tsv_reader = TSVReader('data/example_samples.tsv',
'data/T2D_xls2xml_v1.conf', 'Analysis')
assert not tsv_reader.is_valid()
def test_set_current_conf_key():
# set_current_conf_key() should does nothing
tsv_reader = TSVReader('data/example_samples.tsv',
'data/T2D_xls2xml_v1.conf', 'Sample')
assert tsv_reader.is_valid()
assert set(tsv_reader.get_valid_conf_keys()) == {'Sample'}
tsv_reader.set_current_conf_key('Analysis')
assert tsv_reader.is_valid()
assert set(tsv_reader.get_valid_conf_keys()) == {'Sample'}
tsv_reader = TSVReader('data/example_samples.tsv',
'data/T2D_xls2xml_v1.conf', 'Analysis')
assert not tsv_reader.is_valid()
assert tsv_reader.get_valid_conf_keys() == []
tsv_reader.set_current_conf_key('Sample')
assert not tsv_reader.is_valid()
assert tsv_reader.get_valid_conf_keys() == []