def test_datatable_clear_index(sample_df):
# Test by removing index tag
dt = DataTable(sample_df, index='id')
assert dt.index == 'id'
dt = dt.remove_semantic_tags({'id': 'index'})
assert dt.index is None
assert all(
['index' not in col.semantic_tags for col in dt.columns.values()])
# Test using setter
dt = DataTable(sample_df, index='id')
assert dt.index == 'id'
dt.index = None
assert dt.index is None
assert all(
['index' not in col.semantic_tags for col in dt.columns.values()])
def test_datatable_clear_time_index(sample_df):
# Test by removing time_index tag
dt = DataTable(sample_df, time_index='signup_date')
assert dt.time_index == 'signup_date'
dt = dt.remove_semantic_tags({'signup_date': 'time_index'})
assert dt.time_index is None
assert all(
['time_index' not in col.semantic_tags for col in dt.columns.values()])
# Test using setter
dt = DataTable(sample_df, time_index='signup_date')
assert dt.time_index == 'signup_date'
dt.time_index = None
assert dt.time_index is None
assert all(
['time_index' not in col.semantic_tags for col in dt.columns.values()])
def test_remove_semantic_tags(sample_df):
semantic_tags = {
'full_name': ['tag1', 'tag2', 'tag3'],
'age': ['numeric', 'age'],
'id': ['tag1', 'tag2']
}
dt = DataTable(sample_df,
semantic_tags=semantic_tags,
use_standard_tags=False)
tags_to_remove = {
'full_name': ['tag1', 'tag3'],
'age': 'numeric',
'id': {'tag1'}
}
new_dt = dt.remove_semantic_tags(tags_to_remove)
# Verify original tags were not changed
assert dt.columns['full_name'].semantic_tags == {'tag1', 'tag2', 'tag3'}
assert dt.columns['age'].semantic_tags == {'numeric', 'age'}
assert dt.columns['id'].semantic_tags == {'tag1', 'tag2'}
assert new_dt is not dt
assert new_dt.columns['full_name'].semantic_tags == {'tag2'}
assert new_dt.columns['age'].semantic_tags == {'age'}
assert new_dt.columns['id'].semantic_tags == {'tag2'}
