def test_read_nonmem_dataset(testdata):
path = testdata / 'nonmem' / 'pheno.dta'
colnames = ['ID', 'TIME', 'AMT', 'WGT', 'APGR', 'DV', 'FA1', 'FA2']
df = data.read_nonmem_dataset(path, colnames=colnames, ignore_character='@')
assert list(df.columns) == colnames
assert df.pharmpy.column_type['ID'] == data.ColumnType.ID
assert df['ID'][0] == 1
assert df['TIME'][2] == '12.5' # FIXME! Should be number
raw = data.read_nonmem_dataset(path, colnames=colnames, ignore_character='@', raw=True)
assert raw['ID'][0] == '1'
assert list(df.columns) == colnames
raw2 = data.read_nonmem_dataset(
path, colnames=colnames, ignore_character='@', raw=True, parse_columns=['ID']
)
assert raw2['ID'][0] == 1.0
df_drop = data.read_nonmem_dataset(
path,
colnames=colnames,
ignore_character='@',
drop=[False, False, True, False, False, False, True, False],
)
# FIXME: DROP not decided yet.
# assert list(df_drop.columns) == ['ID', 'TIME', 'WGT', 'APGR', 'DV', 'FA2']
pd.testing.assert_series_equal(df_drop['ID'], df['ID'])
pd.testing.assert_series_equal(df_drop['FA2'], df['FA2'])
def test_nonmem_dataset_with_nonunique_ids():
colnames = ['ID', 'DV']
with pytest.warns(DatasetWarning):
df = data.read_nonmem_dataset(StringIO("1,2\n2,3\n1,4\n2,5"),
colnames=colnames)
assert list(df[df.pharmpy.id_label]) == [1, 2, 3, 4]
assert list(df['DV']) == [2, 3, 4, 5]
def test_nonmem_dataset_with_ignore_accept():
colnames = ['ID', 'DV']
df = data.read_nonmem_dataset(
StringIO("1,2\n1,3\n2,4\n2,5"), colnames=colnames, ignore=['DV.EQN.2']
)
assert len(df) == 3
assert list(df.columns) == colnames
assert list(df.iloc[0]) == [1, 3]
assert list(df.iloc[1]) == [2, 4]
assert list(df.iloc[2]) == [2, 5]
df = data.read_nonmem_dataset(
StringIO("1,2\n1,3\n2,4\n2,a"), colnames=colnames, ignore=['DV.EQ.a', 'DV.EQN.2']
)
assert len(df) == 2
assert list(df.columns) == colnames
assert list(df.iloc[0]) == [1, 3]
assert list(df.iloc[1]) == [2, 4]
with pytest.raises(DatasetError):
df = data.read_nonmem_dataset(
StringIO("1,2\n1,3\n2,4\n2,a"), colnames=colnames, ignore=['DV.EQN.2', 'DV.EQ.a']
)
df = data.read_nonmem_dataset(
StringIO("1,2\n1,3\n2,4\n2,a"), colnames=colnames, ignore=['DV.EQ."a"']
)
assert len(df) == 3
assert list(df.columns) == colnames
assert list(df.iloc[0]) == [1, 2]
assert list(df.iloc[1]) == [1, 3]
assert list(df.iloc[2]) == [2, 4]
df = data.read_nonmem_dataset(
StringIO("1,2\n1,3\n2,4\n2,5"), colnames=colnames, accept=['DV.EQN.2']
)
assert len(df) == 1
assert list(df.columns) == colnames
assert list(df.iloc[0]) == [1, 2]
df = data.read_nonmem_dataset(
StringIO("1,2\n1,3\n2,4\n2,5"), colnames=colnames, ignore=['ID 2']
)
assert len(df) == 2
assert list(df.iloc[0]) == [1, 2]
assert list(df.iloc[1]) == [1, 3]
def test_read_small_nonmem_datasets():
abc = ['A', 'B', 'C']
df = data.read_nonmem_dataset(StringIO("1,2,3"), colnames=abc)
assert list(df.iloc[0]) == [1, 2, 3]
assert list(df.columns) == ['A', 'B', 'C']
df = data.read_nonmem_dataset(StringIO("1, 2 , 3"), colnames=abc)
assert list(df.iloc[0]) == [1, 2, 3]
df = data.read_nonmem_dataset(StringIO("1,,3"), colnames=abc)
assert list(df.iloc[0]) == [1, 0, 3]
df = data.read_nonmem_dataset(StringIO("1,,"), colnames=abc)
assert list(df.iloc[0]) == [1, 0, 0]
df = data.read_nonmem_dataset(StringIO(",2,4"), colnames=abc)
assert list(df.iloc[0]) == [0, 2, 4]
df = data.read_nonmem_dataset(StringIO("1\t2\t3"), colnames=abc)
assert list(df.iloc[0]) == [1, 2, 3]
df = data.read_nonmem_dataset(StringIO("1\t2\t"), colnames=abc)
assert list(df.iloc[0]) == [1, 2, 0]
df = data.read_nonmem_dataset(StringIO("3 4 6"), colnames=abc)
assert list(df.iloc[0]) == [3, 4, 6]
df = data.read_nonmem_dataset(StringIO("3 28 , 341"), colnames=abc)
assert list(df.iloc[0]) == [3, 28, 341]
df = data.read_nonmem_dataset(StringIO(" 1 2 3 "), colnames=abc)
assert list(df.iloc[0]) == [1, 2, 3]
with pytest.raises(DatasetError):
df = data.read_nonmem_dataset(StringIO(" 1 2 3 \n\n"), colnames=abc)
with pytest.raises(DatasetError):
df = data.read_nonmem_dataset(StringIO("1\t2 \t3"), colnames=abc)
# Mismatch length of column_names and data frame
df = data.read_nonmem_dataset(StringIO("1,2,3"), colnames=abc + ['D'])
assert list(df.iloc[0]) == [1, 2, 3, 0]
assert list(df.columns) == ['A', 'B', 'C', 'D']
df = data.read_nonmem_dataset(StringIO("1,2,3,6"), colnames=abc)
assert list(df.iloc[0]) == [1, 2, 3]
assert list(df.columns) == ['A', 'B', 'C']
# Test null_value
df = data.read_nonmem_dataset(StringIO("1,2,"), colnames=abc, null_value=9)
assert list(df.iloc[0]) == [1, 2, 9]