tests/test_verifier.py

from backend.verifier import verify_answer, extract_numbers


SOURCE = (
    "The Consolidated total revenue from operations for the year was "
    "₹ 9,64,693 crore, compared to ₹ 9,01,064 crore in the previous year. "
    "Profit after tax was ₹ 60,812 crore. Attrition was 12.5%."
)
METRICS = {
    "revenue": {"value": 9_64_693 * 10_000_000, "confidence": "high"},
    "profit_after_tax": {"value": 60_812 * 10_000_000, "confidence": "high"},
}


def test_faithful_answer_verifies():
    answer = "Revenue was ₹9,64,693 crore and PAT was ₹60,812 crore. Attrition stood at 12.5%."
    result = verify_answer(answer, SOURCE, METRICS)
    assert result["all_verified"] is True
    assert result["unverified"] == []


def test_invented_numbers_flagged():
    answer = "Revenue was ₹9,64,693 crore. The company spent ₹5,000 crore on R&D and grew 45%."
    result = verify_answer(answer, SOURCE, METRICS)
    assert result["all_verified"] is False
    flagged = {u["value"] for u in result["unverified"]}
    assert 5000.0 in flagged
    assert 45.0 in flagged


def test_unit_conversion_matches_across_formats():
    # $394.3B in the answer vs "394,328 million" in the source
    source = "Total net sales were $394,328 million in fiscal 2022."
    answer = "Apple's revenue was $394.3 billion."
    result = verify_answer(answer, source, None)
    assert result["all_verified"] is True


def test_format_money_output_is_parseable():
    # the exact strings format_money() produces must round-trip
    source = "profit after tax: ₹60,812.0 Cr"
    answer = "PAT was ₹60,812.0 Cr."
    assert verify_answer(answer, source, None)["all_verified"] is True


def test_years_and_small_numbers_ignored():
    answer = "In 2024, the top 3 segments performed well."
    result = verify_answer(answer, "irrelevant source", None)
    assert result["checked"] == 0
    assert result["all_verified"] is True


def test_no_numbers_is_trivially_verified():
    result = verify_answer("Not available in filing.", SOURCE, METRICS)
    assert result["all_verified"] is True
    assert result["checked"] == 0


def test_extract_numbers_shapes():
    nums = extract_numbers("Revenue was ₹1,22,670.1 crore, margin 24.6%.")
    raws = [n["raw"] for n in nums]
    assert any("crore" in r for r in raws)
    assert any("%" in r for r in raws)