import sys, os
sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..")))
from comparator.aligner import align_records


def test_align_by_key():
    c = [{"CUST-ID": "C001", "AMT": 100}, {"CUST-ID": "C002", "AMT": 200}]
    s = [{"CUST-ID": "C002", "AMT": 200}, {"CUST-ID": "C001", "AMT": 100}]
    result = align_records(c, s, key_field="CUST-ID")
    assert len(result) == 2
    assert all(st == "MATCHED" for _, _, st in result)


def test_missing_in_spark():
    c = [{"CUST-ID": "C001"}, {"CUST-ID": "C002"}]
    s = [{"CUST-ID": "C001"}]
    result = align_records(c, s, key_field="CUST-ID")
    assert "MISSING_IN_SPARK" in [st for _, _, st in result]


def test_extra_in_spark():
    c = [{"CUST-ID": "C001"}]
    s = [{"CUST-ID": "C001"}, {"CUST-ID": "C002"}]
    result = align_records(c, s, key_field="CUST-ID")
    assert "EXTRA_IN_SPARK" in [st for _, _, st in result]


def test_empty_inputs():
    assert align_records([], [], "key") == []


def test_duplicate_keys():
    c = [{"ID": "K1", "V": 1}, {"ID": "K1", "V": 2}]
    s = [{"ID": "K1", "V": 1}, {"ID": "K1", "V": 2}]
    assert len(align_records(c, s, key_field="ID")) == 2


def test_none_key():
    assert len(align_records([{"ID": None, "V": 1}], [{"ID": None, "V": 1}], "ID")) == 1