def test_translate_flowsheet_id_to_fid(internal_id, fid):
df = pd.DataFrame([
{
'FlowsheetRowID': internal_id,
'ExtraColumn': 'ExtraValue'
},
])
df_correct = pd.DataFrame([
{
'fid': fid,
'FlowsheetRowID': internal_id,
'ExtraColumn': 'ExtraValue'
},
])
result = translate.translate_epic_id_to_fid(df=df.copy(),
col='FlowsheetRowID',
new_col='fid',
config_map=flowsheet_ids)
assert test_utils.dataframe_equality(result, df_correct)
df_correct_drop = pd.DataFrame([
{
'fid': fid,
'ExtraColumn': 'ExtraValue'
},
])
result_drop = translate.translate_epic_id_to_fid(df=df.copy(),
col='FlowsheetRowID',
new_col='fid',
config_map=flowsheet_ids,
drop_original=True)
assert test_utils.dataframe_equality(result_drop, df_correct_drop)
def test_unlistify_empty_list():
df = pd.DataFrame([
{
'A': 'A_0_1',
'B': ['B_0_1', 'B_0_2'],
'C': 'C_0_1',
},
{
'A': 'A_1_1',
'B': [],
'C': 'C_1_1',
},
])
df_extracted = pd.DataFrame([
{
'A': 'A_0_1',
'B': 'B_0_1',
'C': 'C_0_1',
},
{
'A': 'A_0_1',
'B': 'B_0_2',
'C': 'C_0_1',
},
{
'A': 'A_1_1',
'B': np.nan,
'C': 'C_1_1',
},
])
result = pandas_utils.unlistify_pandas_column(df.copy(), 'B')
assert test_utils.dataframe_equality(df_extracted, result)
def test_unlistify_lists_of_length_one():
df = pd.DataFrame([
{
'A': 'A_0_1',
'B': ['B_0_1'],
'C': 'C_0_1',
},
{
'A': 'A_1_1',
'B': ['B_1_1'],
'C': 'C_1_1',
},
])
df_extracted = pd.DataFrame([
{
'A': 'A_0_1',
'B': 'B_0_1',
'C': 'C_0_1',
},
{
'A': 'A_1_1',
'B': 'B_1_1',
'C': 'C_1_1',
},
])
result = pandas_utils.unlistify_pandas_column(df.copy(), 'B')
assert test_utils.dataframe_equality(df_extracted, result)
def test_unlistify_multiple_lists():
df = pd.DataFrame([
{
'A': 'A_0_1',
'B': ['B_0_1', 'B_0_2', 'B_0_3'],
'C': ['C_0_1'],
},
{
'A': 'A_1_1',
'B': ['B_1_1', 'B_1_2'],
'C': ['C_1_1', 'C_1_2'],
},
])
df_extracted = pd.DataFrame([
{
'A': 'A_0_1',
'B': 'B_0_1',
'C': 'C_0_1',
},
{
'A': 'A_0_1',
'B': 'B_0_2',
'C': 'C_0_1',
},
{
'A': 'A_0_1',
'B': 'B_0_3',
'C': 'C_0_1',
},
{
'A': 'A_1_1',
'B': 'B_1_1',
'C': 'C_1_1',
},
{
'A': 'A_1_1',
'B': 'B_1_2',
'C': 'C_1_1',
},
{
'A': 'A_1_1',
'B': 'B_1_1',
'C': 'C_1_2',
},
{
'A': 'A_1_1',
'B': 'B_1_2',
'C': 'C_1_2',
},
])
result_1 = pandas_utils.unlistify_pandas_column(df.copy(), 'C')
result_2 = pandas_utils.unlistify_pandas_column(result_1, 'B')
assert test_utils.dataframe_equality(df_extracted, result_2)
def test_unlistify_lists_of_same_lengths():
df = pd.DataFrame([{
'A': 'A_' + str(j) + '_1',
'B': ['B_' + str(j) + '_' + str(i) for i in range(3)],
'C': 'C_' + str(j) + '_1',
} for j in range(6)])
df_extracted = pd.DataFrame([{
'A': 'A_' + str(int(j / 3)) + '_1',
'B': 'B_' + str(int(j / 3)) + '_' + str(j % 3),
'C': 'C_' + str(int(j / 3)) + '_1',
} for j in range(18)])
result = pandas_utils.unlistify_pandas_column(df.copy(), 'B')
assert test_utils.dataframe_equality(df_extracted, result)
def test_unlist():
empty_list = []
df = pd.DataFrame([{
'ComponentID': 'fid_' + str(k),
'Value': [str(i) for i in range(2)],
'tsp': str(j),
'Unit': 'unit_' + str(k),
} for k in range(2) for j in range(10)])
df_extracted = pd.DataFrame([{
'ComponentID': 'fid_' + str(k),
'Value': str(j % 2),
'tsp': str(int(j / 2)),
'Unit': 'unit_' + str(k),
} for k in range(2) for j in range(20)])
result = restructure.unlist(df.copy(), 'Value')
assert test_utils.dataframe_equality(df_extracted, result)
def test_translate_lab_result_to_fids(component_id, fid):
df = pd.DataFrame([
{
'ComponentID': component_id,
'ExtraColumn': 'ExtraValue'
},
])
df_correct = pd.DataFrame([
{
'fid': fid,
'ComponentID': component_id,
'ExtraColumn': 'ExtraValue'
},
])
result = translate.translate_epic_id_to_fid(df=df.copy(),
col='ComponentID',
new_col='fid',
config_map=component_ids)
assert test_utils.dataframe_equality(result, df_correct)
def test_extract():
df = pd.DataFrame([{
'fid': 'fid_' + str(j),
'dict_col': {
'RawValue': 'val_' + str(j),
'Instant': 'tsp_' + str(j),
'Unneeded': 'fid_' + str(j),
},
'Unit': 'unit_' + str(j),
} for j in range(10)])
df_extracted = pd.DataFrame([{
'fid': 'fid_' + str(j),
'new_raw_value': 'val_' + str(j),
'tsp': 'tsp_' + str(j),
'Unit': 'unit_' + str(j),
} for j in range(10)])
result = restructure.extract(df.copy(), 'dict_col', {
'RawValue': 'new_raw_value',
'Instant': 'tsp',
})
assert test_utils.dataframe_equality(df_extracted, result)
def test_extract_internal_ids():
df = pd.DataFrame([{
'ID_col': [{
'Type': 'bad_type1',
}, {
'Type': 'good_type',
'ID': j
}, {
'Type': 'bad_type2',
'ID': str(j + 10)
}],
'Extra':
'still_here',
} for j in range(5)])
df_correct = pd.DataFrame([{
'ID_col': str(j),
'Extra': 'still_here'
} for j in range(5)])
result = restructure.extract_id_from_list(df.copy(), 'ID_col', 'good_type')
assert test_utils.dataframe_equality(df_correct, result)