def test_match_columns(self):
raw = [
{
'first name': 'Mary',
'LASTNAME': 'Nichols',
'Middle__Name': 'D'
},
{
'first name': 'Lucy',
'LASTNAME': 'Peterson',
'Middle__Name': 'S'
},
]
tbl = Table(raw)
desired_raw = [
{
'first_name': 'Mary',
'middle_name': 'D',
'last_name': 'Nichols'
},
{
'first_name': 'Lucy',
'middle_name': 'S',
'last_name': 'Peterson'
},
]
desired_tbl = Table(desired_raw)
# Test with fuzzy matching
tbl.match_columns(desired_tbl.columns)
assert_matching_tables(desired_tbl, tbl)
# Test disable fuzzy matching, and fail due due to the missing cols
self.assertRaises(TypeError,
Table(raw).match_columns,
desired_tbl.columns,
fuzzy_match=False,
if_missing_columns='fail')
# Test disable fuzzy matching, and fail due to the extra cols
self.assertRaises(TypeError,
Table(raw).match_columns,
desired_tbl.columns,
fuzzy_match=False,
if_extra_columns='fail')
# Test table that already has the right columns, shouldn't need fuzzy match
tbl = Table(desired_raw)
tbl.match_columns(desired_tbl.columns,
fuzzy_match=False,
if_missing_columns='fail',
if_extra_columns='fail')
assert_matching_tables(desired_tbl, tbl)
# Test table with missing col, verify the missing col gets added by default
tbl = Table([
{
'first name': 'Mary',
'LASTNAME': 'Nichols'
},
{
'first name': 'Lucy',
'LASTNAME': 'Peterson'
},
])
tbl.match_columns(desired_tbl.columns)
desired_tbl = (
Table(desired_raw).remove_column('middle_name').add_column(
'middle_name', index=1))
assert_matching_tables(desired_tbl, tbl)
# Test table with extra col, verify the extra col gets removed by default
tbl = Table([
{
'first name': 'Mary',
'LASTNAME': 'Nichols',
'Age': 32,
'Middle__Name': 'D'
},
{
'first name': 'Lucy',
'LASTNAME': 'Peterson',
'Age': 26,
'Middle__Name': 'S'
},
])
desired_tbl = Table(desired_raw)
tbl.match_columns(desired_tbl.columns)
assert_matching_tables(desired_tbl, tbl)
# Test table with two columns that normalize the same and aren't in desired cols, verify
# they both get removed.
tbl = Table([
{
'first name': 'Mary',
'LASTNAME': 'Nichols',
'Age': 32,
'Middle__Name': 'D',
'AGE': None
},
{
'first name': 'Lucy',
'LASTNAME': 'Peterson',
'Age': 26,
'Middle__Name': 'S',
'AGE': None
},
])
tbl.match_columns(desired_tbl.columns)
assert_matching_tables(desired_tbl, tbl)
# Test table with two columns that match desired cols, verify only the first gets kept.
tbl = Table([
{
'first name': 'Mary',
'LASTNAME': 'Nichols',
'First Name': None,
'Middle__Name': 'D'
},
{
'first name': 'Lucy',
'LASTNAME': 'Peterson',
'First Name': None,
'Middle__Name': 'S'
},
])
tbl.match_columns(desired_tbl.columns)
assert_matching_tables(desired_tbl, tbl)
