conservancy_beancount/tests/test_reconcile.py

import datetime
import decimal

from conservancy_beancount.reconcile.prototype_amex_reconciler import (
    match_statement_and_books,
    remove_payee_junk,
    date_proximity,
    remove_duplicate_words,
    payee_match,
)

S1 = {
    'date': datetime.date(2022, 1, 1),
    'amount': decimal.Decimal('10.00'),
    'payee': 'Patreon         / Patreon   / 123456/ ST-A1B2C3D4G5H6       /',
    'check_id': '',
    'line': 222,
}
S2 = {
    'date': datetime.date(2022, 1, 2),
    'amount': decimal.Decimal('20.00'),
    'payee': 'BT*LINODE           PHILADELPHIA        P',
    'check_id': '',
    'line': 333,
}
S3 = {
    'date': datetime.date(2022, 1, 3),
    'amount': decimal.Decimal('30.00'),
    'payee': 'USPS PO 4067540039 0PORTLAND            OR',
    'check_id': '',
    'line': 444,
}

B1 = {
    'date': datetime.date(2022, 1, 1),
    'amount': decimal.Decimal('10.00'),
    'payee': 'Patreon',
    'check_id': '',
    'filename': '2022/imports.beancount',
    'line': 777,
    'bank_statement': "Financial/Bank-Statements/AMEX/2022-01-12_AMEX_statement.pdf"
}
B2 = {
    'date': datetime.date(2022, 1, 2),
    'amount': decimal.Decimal('20.00'),
    'payee': 'Linode',
    'check_id': '',
    'filename': '2022/main.beancount',
    'line': 888,
    'bank_statement': "Financial/Bank-Statements/AMEX/2022-01-12_AMEX_statement.pdf"
}
B3_next_day = {
    'date': datetime.date(2022, 1, 4),
    'amount': decimal.Decimal('30.00'),
    'payee': 'USPS',
    'check_id': '',
    'filename': '2022/main.beancount',
    'line': 999,
    'bank_statement': "Financial/Bank-Statements/AMEX/2022-01-12_AMEX_statement.pdf"
}
B3_next_week = {
    'date': datetime.date(2022, 1, 10),
    'amount': decimal.Decimal('30.00'),
    'payee': 'USPS',
    'check_id': '',
    'filename': '2022/main.beancount',
    'line': 999,
    'bank_statement': "Financial/Bank-Statements/AMEX/2022-01-12_AMEX_statement.pdf"
}
B3_mismatch_amount = {
    'date': datetime.date(2022, 1, 3),
    'amount': decimal.Decimal('31.00'),
    'payee': 'USPS',
    'check_id': '',
    'filename': '2022/main.beancount',
    'line': 999,
    'bank_statement': "Financial/Bank-Statements/AMEX/2022-01-12_AMEX_statement.pdf"
}
B3_payee_mismatch_1 = {
    'date': datetime.date(2022, 1, 3),
    'amount': decimal.Decimal('30.00'),
    'payee': 'Credit X',
    'check_id': '',
    'filename': '2022/main.beancount',
    'line': 999,
    'bank_statement': "Financial/Bank-Statements/AMEX/2022-01-12_AMEX_statement.pdf"
}
B3_payee_mismatch_2 = {
    'date': datetime.date(2022, 1, 3),
    'amount': decimal.Decimal('30.00'),
    'payee': 'Credit Y',
    'check_id': '',
    'filename': '2022/main.beancount',
    'line': 999,
    'bank_statement': "Financial/Bank-Statements/AMEX/2022-01-12_AMEX_statement.pdf"
}


def test_one_exact_match():
    statement = [S1]
    books = [B1]
    assert match_statement_and_books(statement, books) == [
        ([S1], [B1], []),
    ]

def test_multiple_exact_matches():
    statement = [S1, S2]
    books = [B1, B2]
    assert match_statement_and_books(statement, books) == [
        ([S1], [B1], []),
        ([S2], [B2], []),
    ]

def test_one_mismatch():
    statement = [S1]
    books = []
    assert match_statement_and_books(statement, books) == [
        ([S1], [], ['no match']),
    ]

def test_multiple_mismatches():
    statement = [S1]
    books = [B2]
    assert match_statement_and_books(statement, books) == [
        ([S1], [], ['no match']),
        ([], [B2], ['no match']),
    ]

def test_next_day_matches():
    statement = [S3]
    books = [B3_next_day]
    assert match_statement_and_books(statement, books) == [
        ([S3], [B3_next_day], ['+/- 1 days']),
    ]

def test_next_week_matches():
    statement = [S3]
    books = [B3_next_week]
    assert match_statement_and_books(statement, books) == [
        ([S3], [B3_next_week], ['+/- 7 days']),
    ]

def test_incorrect_amount_does_not_match():
    statement = [S3]
    books = [B3_mismatch_amount]
    assert match_statement_and_books(statement, books) == [
        ([S3], [], ['no match']),
        ([], [B3_mismatch_amount], ['no match']),
    ]

def test_payee_mismatch_ok_when_only_one_that_amount_and_date():
    statement = [S3]
    books = [B3_payee_mismatch_1]
    assert match_statement_and_books(statement, books) == [
        ([S3], [B3_payee_mismatch_1], ['payee mismatch', 'only one decent match']),
    ]

def test_payee_mismatch_not_ok_when_multiple_that_amount_and_date():
    statement = [S3]
    books = [B3_payee_mismatch_1, B3_payee_mismatch_2]
    match = match_statement_and_books(statement, books)
    assert match == [
        ([S3], [], ['no match']),
        ([], [B3_payee_mismatch_1], ['no match']),
        ([], [B3_payee_mismatch_2], ['no match']),
    ]

# def test_subset_sum_with_same_date_and_payee():

def test_remove_payee_junk():
    assert remove_payee_junk('WIDGETSRUS INC PAYMENT 1') == 'WIDGETSRUS'
    assert remove_payee_junk('0000010017') == '10017'

def test_date_proximity():
    assert date_proximity(datetime.date(2021, 8, 23), datetime.date(2021, 8, 23)) == 1.0
    assert date_proximity(datetime.date(2021, 8, 23), datetime.date(2021, 8, 23) - datetime.timedelta(days=30)) == 0.5
    assert date_proximity(datetime.date(2021, 8, 23), datetime.date(2021, 8, 23) - datetime.timedelta(days=60)) == 0.0


def test_remove_duplicate_words():
    assert remove_duplicate_words('Hi Foo Kow FOO') == 'Hi Foo Kow'

def test_remove_duplicate_words():
    assert remove_duplicate_words('Hi Foo Kow FOO') == 'Hi Foo Kow'

def test_payee_matches_when_first_word_matches():
    assert payee_match('Gandi San Francisco', 'Gandi example.com renewal 1234567') == 1.0
    assert payee_match('USPS 123456789 Portland', 'USPS John Brown') == 0.8