def test_dataset_modified(self):
invalid = datetime(2100, 1, 1)
ds = xport.Dataset(modified=invalid)
assert ds.modified == invalid
with pytest.raises(ValueError):
xport.v56.dumps(ds)
with pytest.raises((TypeError, AttributeError)):
ds = xport.Dataset(modified=1)
xport.v56.dumps(ds)
def test_dataset_created(self):
invalid = datetime(1800, 1, 1)
ds = xport.Dataset(created=invalid)
assert ds.created == invalid
with pytest.raises(ValueError):
xport.v56.dumps(ds)
with pytest.raises((TypeError, AttributeError)):
ds = xport.Dataset(created='2000-Jan-01')
xport.v56.dumps(ds)
def test_init(self):
"""
Verify initialization.
"""
v = xport.Dataset()
for name in v._metadata:
getattr(v, name) # Does not raise an error.
def test_troublesome_text(self):
"""
Some text patterns have been trouble in the past.
"""
trouble = xport.Variable(["'<>"], dtype='string')
dataset = xport.Dataset({'a': trouble}, name='trouble')
library = xport.Library(dataset)
with pytest.warns(UserWarning, match=r'Converting column dtypes'):
assert self.dump_and_load(library) == library
def test_dumps_name_and_label_length_validation(self):
"""
Verify variable and dataset name and label length.
"""
# Names must be <= 8 characters.
# Labels must be <= 40 characters.
# SAS v8 Transport Files allow longer labels.
invalid = [
xport.Library(xport.Dataset(), sas_version='a' * 9),
xport.Library(xport.Dataset(name='a' * 9)),
xport.Library(xport.Dataset(label='a' * 41)),
xport.Library(xport.Dataset({'a' * 9: [1.0]})),
xport.Library(
xport.Dataset({'a': xport.Variable([1.0], label='a' * 41)})),
]
for bad_metadata in invalid:
with pytest.raises(ValueError):
xport.v56.dumps(bad_metadata)
def test_invalid_values(self):
invalid = [
'\N{snowman}',
]
for bad in invalid:
library = xport.Library(xport.Dataset({'a': [bad]}))
with pytest.raises(ValueError):
with pytest.warns(UserWarning,
match=r'Converting column dtypes'):
xport.v56.dumps(library)
def test_decode(library, library_bytestring):
"""
Verify the command line executable can decode a library.
"""
cmd = 'python -m xport -'
argv = cmd.split()
proc = subprocess.run(argv, capture_output=True, input=library_bytestring)
fp = StringIO(proc.stdout.decode())
df = pd.read_csv(fp)
ds = xport.Dataset(df)
assert (ds == next(iter(library.values()))).all(axis=None)
def test_output_file(library, library_bytestring, tmp_path):
"""
Verify CLI can write output to a file.
"""
filepath = tmp_path / 'tmp.csv'
cmd = f'python -m xport - {filepath}'
argv = cmd.split()
subprocess.run(argv, capture_output=True, input=library_bytestring)
with open(filepath) as f:
df = pd.read_csv(f)
ds = xport.Dataset(df)
assert (ds == next(iter(library.values()))).all(axis=None)
def test_invalid_types(self):
"""
Verify invalid types raise errors on write.
"""
invalid = [
b'\x00',
object(),
(1, 2, 3),
]
for bad in invalid:
with pytest.warns(UserWarning, match=r'Converting column dtypes'):
with pytest.raises(TypeError):
library = xport.Library(xport.Dataset({'a': [bad]}))
xport.v56.dumps(library)
def test_numeric_type_conversion(self):
"""
Verify numeric types convert to float when writing.
"""
numerics = [
1,
True,
]
for x in numerics:
with pytest.warns(UserWarning, match=r'Converting column dtypes'):
library = xport.Library({'A': xport.Dataset({'x': [x]})})
output = self.dump_and_load(library)
assert output['A']['x'].dtype.name == 'float64'
assert output['A']['x'].iloc[0] == 1.0
def test_contents(self):
"""
Verify variables metadata summary.
"""
ds = xport.Dataset(
data={
'a': [1],
'b': xport.Variable(['x'], label='Beta'),
'c': [None],
},
name='EXAMPLE',
label='Example',
)
ds['a'].vtype = xport.VariableType.NUMERIC
ds['b'].vtype = xport.VariableType.CHARACTER
got = ds.contents
assert list(got.index) == [1, 2, 3]
assert list(got['Label']) == ['', 'Beta', '']
assert list(got['Type']) == ['Numeric', 'Character', '']
def library():
"""
Create a 4-column, 6-row dataset with numbers and text.
"""
ds = xport.Dataset(
data={
'VIT_STAT': ['ALIVE'] * 3 + ['DEAD'] * 3,
'ECON': ['POOR', 'NOT', 'UNK'] * 2,
'COUNT': [1216, 1761, 2517, 254, 60, 137],
'TEMP': [98.6, 95.4, 86.7, 93.4, 103.5, 56.7],
},
name='ECON',
label='Blank-padded dataset label',
dataset_type='',
)
ds.created = ds.modified = datetime(2015, 11, 13, 10, 35, 8)
ds.sas_os = 'W32_7PRO'
ds.sas_version = '9.3'
ds['VIT_STAT'].label = 'Vital status'
ds['VIT_STAT'].format = '$5.'
ds['VIT_STAT'].informat = xport.Informat()
ds['VIT_STAT'].width = 8
ds['ECON'].label = 'Economic status'
ds['ECON'].format = xport.Format('$CHAR', 4, 0,
xport.FormatAlignment.RIGHT)
ds['ECON'].informat = xport.Informat()
ds['ECON'].width = 8
ds['COUNT'].label = 'Count'
ds['COUNT'].format = 'comma8.0'
ds['COUNT'].informat = xport.Informat()
ds['COUNT'].width = 8
ds['TEMP'].label = 'Temperature'
ds['TEMP'].format = '8.1'
ds['TEMP'].informat = xport.Informat()
ds['TEMP'].width = 8
return xport.Library(
members=[ds],
created=ds.created,
modified=ds.modified,
sas_os=ds.sas_os,
sas_version=ds.sas_version,
)
def test_copy_metadata(self):
"""
Verify ``DataFrame`` methods that copy will keep SAS metadata.
"""
ds = xport.Dataset(
data={
'a': [1],
'b': xport.Variable(['x'], label='Beta')
},
name='EXAMPLE',
label='Example',
)
self.compare_metadata(ds.copy(), ds)
self.compare_metadata(
ds.append(pd.DataFrame({
'a': [2],
'b': ['y'],
})),
ds,
)
self.compare_metadata(pd.concat([ds, ds]), ds)
# Insert rows to DataFrame:
TD_new_row = {'Row': var_counter_rows, 'STUDYID': CT_STUDYID, 'DOMAIN': var_domain_<..>, ...}
TD_df = TV_df.append(TD_new_row, ignore_index=True)
#
var_counter_<..> += 1
var_counter_rows += 1
"""
#
# = Generate SAS files: = =
# Source: https://github.com/selik/xport
# The SAS Transport (XPORT) format only supports two kinds of data. Each value is either numeric or character, so xport.load decodes the values as either str or float.
#
# TA:
TA_ds = xport.Dataset(TA_df, name='TA', label='Trial Arms (TA) data')
# SAS variable names are limited to 8 characters. As with Pandas dataframes, you must change the name on the dataset rather than the column directly.
TA_ds = TA_ds.rename(columns={k: k.upper()[:8] for k in TA_ds})
# Libraries can have multiple datasets.
TA_library = xport.Library({'TA': TA_ds})
#
with open('TA.xpt', 'wb') as f:
xport.v56.dump(TA_library, f)
#
# TE:
TE_ds = xport.Dataset(TE_df, name='TE', label='Trial Elements (TE) data')
# SAS variable names are limited to 8 characters. As with Pandas dataframes, you must change the name on the dataset rather than the column directly.
TE_ds = TE_ds.rename(columns={k: k.upper()[:8] for k in TE_ds})
# Libraries can have multiple datasets.
TE_library = xport.Library({'TE': TE_ds})
#
def test_create_from_list(self):
lib = xport.Library(xport.Dataset())
assert None in lib
with pytest.warns(UserWarning, match=r'More than one dataset named'):
xport.Library([xport.Dataset(), xport.Dataset()])
def test_create_from_mapping(self):
with pytest.warns(UserWarning, match=r'Set dataset name'):
lib = xport.Library({'x': xport.Dataset()})
assert 'x' in lib
with pytest.raises(ValueError):
xport.Library({'x': xport.Dataset(name='y')})
