def test_init_args_progressbar_default():
test_args = ['etude.py', '--no-metrics']
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert args.progressbar_output == 'stderr'
assert not args.progressbar_disabled
assert args.progressbar_file == sys.stderr
def test_count_ref_set_csv_out():
presaved_file = 'tests/data/i2b2_2016_track-1_csv_out.csv'
try:
tmp_descriptor, tmp_file = tempfile.mkstemp()
os.close(tmp_descriptor)
command_line_args = [
'etude.py', '--reference-input',
'tests/data/i2b2_2016_track-1_reference', '--reference-config',
'config/i2b2_2016_track-1.conf', '--csv-out', tmp_file,
'--by-file', '--by-type', '--by-file-and-type',
'--by-type-and-file', '--print-counts', '--no-metrics'
]
with patch.object(sys, 'argv', command_line_args):
args = etude.init_args()
namespaces , document_data , patterns = \
args_and_configs.process_config( config_file = args.reference_config ,
score_key = args.score_key ,
score_values = args.score_values )
import csv
with open(presaved_file, 'r') as fp:
reloaded_csv = fp.read()
etude.count_ref_set(this_ns=namespaces,
this_dd=document_data,
this_patterns=patterns,
this_folder=args.reference_input,
args=args,
file_prefix=args.file_prefix,
file_suffix=args.file_suffix[0])
with open(tmp_file, 'r') as fp:
new_csv = fp.read()
assert new_csv == reloaded_csv
finally:
os.remove(tmp_file)
def test_init_args_file_suffix_duplex():
test_args = ['etude.py', '--no-metrics', '--file-suffix', '.xyz', '.abc']
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert len(args.file_suffix) == 2
assert args.file_suffix[0] == '.xyz'
assert args.file_suffix[1] == '.abc'
def test_count_ref_set_default(capsys):
presaved_file = 'tests/data/i2b2_2016_track-1_reference_out/0005_gs.xml'
command_line_args = [
'etude.py', '--reference-input',
'tests/data/i2b2_2016_track-1_reference', '--reference-config',
'config/i2b2_2016_track-1.conf', '--print-counts', '--no-metrics'
]
with patch.object(sys, 'argv', command_line_args):
args = etude.init_args()
namespaces , document_data , patterns = \
args_and_configs.process_config( config_file = args.reference_config ,
score_key = args.score_key ,
score_values = args.score_values )
with open(presaved_file, 'r') as fp:
reloaded_json = json.load(fp)
etude.count_ref_set(this_ns=namespaces,
this_dd=document_data,
this_patterns=patterns,
this_folder=args.reference_input,
args=args,
file_prefix=args.file_prefix,
file_suffix=args.file_suffix[0])
default_out, err = capsys.readouterr()
expected_values = [['counts', 'n'], ['Total', '482']]
for expected_values in expected_values:
print(args.delim.join('{}'.format(m) for m in expected_values))
expected_out, err = capsys.readouterr()
default_out = default_out.strip()
expected_out = expected_out.strip()
assert default_out == expected_out
def test_init_args_progressbar_none():
test_args = ['etude.py', '--no-metrics', '--progressbar-output', 'none']
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert args.progressbar_output == 'none'
assert args.progressbar_disabled
assert args.progressbar_file is None
def test_init_args_corpus_out():
test_args = [
'etude.py', '--reference-input', '/tmp/reference', '--test-input',
'/tmp/test', '--corpus-out', '/tmp/corpusOut'
]
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert args.corpus_out == '/tmp/corpusOut'
def test_init_args_verbose():
test_args = [
'etude.py', '--reference-input', '/tmp/reference', '--test-input',
'/tmp/test', '--verbose'
]
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert args.verbose == True
def test_init_args_default_f_beta_values():
test_args = [
'etude.py', '--reference-input', '/tmp/reference', '--test-input',
'/tmp/test'
]
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert args.f_beta_values == []
def test_init_args_skip_chars_and_whitespace_flag():
test_args = [
'etude.py', '--reference-input', '/tmp/reference', '--test-input',
'/tmp/test', '--ignore-whitespace', '--skip-chars', r'[z\|]'
]
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert args.ignore_whitespace == True
assert args.skip_chars == r'[z\|]'
def test_init_args_heed_whitespace():
test_args = [
'etude.py', '--reference-input', '/tmp/reference', '--test-input',
'/tmp/test', '--heed-whitespace'
]
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert args.ignore_whitespace == False
assert args.skip_chars == None
def test_init_args_f1_and_1():
test_args = [
'etude.py', '--reference-input', '/tmp/reference', '--test-input',
'/tmp/test', '-m', 'Precision', 'F1', 'Recall', '--f-beta-values', '1'
]
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert 'F' not in args.metrics_list
assert ['Precision', 'F1', 'Recall'] == args.metrics_list
assert args.f_beta_values == ['1']
def test_default_init_args():
test_args = [
'etude.py', '--reference-input', '/tmp/reference', '--test-input',
'/tmp/test'
]
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert args.reference_input == '/tmp/reference'
assert args.test_input == '/tmp/test'
assert args.corpus_out == None
assert args.verbose == False
def test_init_args_multiple_f_beta_values():
test_args = [
'etude.py', '--reference-input', '/tmp/reference', '--test-input',
'/tmp/test', '-m', 'Precision', 'F', 'Recall', '--f-beta-values', '4',
'0.34'
]
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert 'F' not in args.metrics_list
assert ['Precision', 'F4', 'F0.34', 'Recall'] == args.metrics_list
## Note: the beta values are reversed because of how we need
## to insert them into the metrics_list
assert args.f_beta_values == ['0.34', '4']
def test_count_ref_set_by_type_and_file(capsys):
presaved_file = 'tests/data/i2b2_2016_track-1_reference_out/0005_gs.xml'
command_line_args = [
'etude.py', '--reference-input',
'tests/data/i2b2_2016_track-1_reference', '--reference-config',
'config/i2b2_2016_track-1.conf', '--by-type', '--by-file',
'--print-counts', '--no-metrics'
]
with patch.object(sys, 'argv', command_line_args):
args = etude.init_args()
namespaces , document_data , patterns = \
args_and_configs.process_config( config_file = args.reference_config ,
score_key = args.score_key ,
score_values = args.score_values )
with open(presaved_file, 'r') as fp:
reloaded_json = json.load(fp)
etude.count_ref_set(this_ns=namespaces,
this_dd=document_data,
this_patterns=patterns,
this_folder=args.reference_input,
args=args,
file_prefix=args.file_prefix,
file_suffix=args.file_suffix[0])
default_out, err = capsys.readouterr()
expected_values = [['counts', 'n'], ['Total', '482'],
['0005_gs.xml', '36'], ['0016_gs.xml', '54'],
['0267_gs.xml', '63'], ['0273_gs.xml', '35'],
['0389_gs.xml', '40'], ['0475_gs.xml', '46'],
['0617_gs.xml', '38'], ['0709_gs.xml', '45'],
['0982_gs.xml', '100'], ['0992_gs.xml', '25'],
['Age', '92'], ['DateTime',
'124'], ['HCUnit', '76'],
['OtherGeo', '5'], ['OtherID', '7'],
['OtherOrg', '21'], ['Patient', '19'],
['PhoneFax', '6'], ['Provider', '64'], ['SSN', '0'],
['StateCountry', '33'], ['StreetCity', '29'],
['Zip', '4'], ['eAddress', '2']]
for expected_values in expected_values:
print(args.delim.join('{}'.format(m) for m in expected_values))
expected_out, err = capsys.readouterr()
default_out = default_out.strip()
expected_out = expected_out.strip()
assert default_out == expected_out
def test_init_args_file_suffix_simplex_dash():
test_args = ['etude.py', '--no-metrics', '--file-suffix', ' -03.xyz']
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert len(args.file_suffix) == 1
assert args.file_suffix[0] == '-03.xyz'
def test_init_args_file_prefix_simplex():
test_args = ['etude.py', '--no-metrics', '--file-prefix', '.xyz']
with patch.object(sys, 'argv', test_args):
args = etude.init_args()
assert args.file_prefix == '.xyz'
