personal-data/test/test_deduplicate.py

import datetime
from decimal import Decimal

from frozendict import frozendict

from personal_data.data import DeduplicateMode
from personal_data.main import deduplicate_dicts

LIST = [
    frozendict({'a': 0, 'b': 12, 't': 300}),
    frozendict({'a': 0, 'b': 12, 't': 301}),
    frozendict({'a': 1, 'b': 2, 't': 300}),
    frozendict({'a': 1, 'b': 2, 't': 301}),
    frozendict({'a': 1, 'b': 2, 't': 302}),
    frozendict({'a': 1, 'b': 2, 't': 303}),
]


def test_no_fields_to_ignore():
    for mode in DeduplicateMode:
        ls, fields = deduplicate_dicts(LIST, mode, [])
        assert fields == ['a', 'b', 't']
        assert ls == LIST


def test_only_latest():
    ls, fields = deduplicate_dicts(LIST, DeduplicateMode.ONLY_LATEST, ['t'])
    assert fields == ['a', 'b', 't']
    assert ls == LIST[:3]


def test_all_fields():
    ls, fields = deduplicate_dicts(LIST, DeduplicateMode.BY_ALL_COLUMNS, ['t'])
    assert fields == ['a', 'b', 't']
    print(ls)
    assert ls == [
        frozendict({'a': 0, 'b': 12, 't': 300}),
        frozendict({'a': 1, 'b': 2, 't': 300}),
    ]


def test_all_fields():
    ls, fields = deduplicate_dicts(LIST + LIST, DeduplicateMode.BY_ALL_COLUMNS, ['t'])
    assert fields == ['a', 'b', 't']
    print(ls)
    assert ls == [
        frozendict({'a': 0, 'b': 12, 't': 300}),
        frozendict({'a': 1, 'b': 2, 't': 300}),
    ]


LIST_2 = [
    frozendict(
        {
            'weight.sample_time': datetime.datetime(
                2024,
                5,
                28,
                6,
                27,
                31,
                134506,
                tzinfo=datetime.timezone.utc,
            ),
            'weight.kg': Decimal('73.6'),
        },
    ),
    frozendict(
        {
            'weight.sample_time': datetime.datetime(
                2024,
                6,
                1,
                7,
                36,
                9,
                590355,
                tzinfo=datetime.timezone.utc,
            ),
            'weight.kg': Decimal('74.7'),
        },
    ),
]


def test_deduplicate_weight():
    ls, fields = deduplicate_dicts(LIST_2, DeduplicateMode.BY_ALL_COLUMNS, [])
    assert fields == ['weight.sample_time', 'weight.kg']
    assert ls == LIST_2


def test_deduplicate_weight_2():
    ls, fields = deduplicate_dicts(LIST_2 + LIST_2, DeduplicateMode.BY_ALL_COLUMNS, [])
    assert fields == ['weight.sample_time', 'weight.kg']
    assert ls == LIST_2