tests/test_metrics_extractor.py

from backend.metrics_extractor import (
    extract_metrics,
    find_metric_in_text,
    normalize_currency_symbols,
    normalize_grouped_digits,
    parse_value,
    score_confidence,
    demote_confidence,
)


HDFC_TEXT = """
HDFC Bank reported total income of ₹250,000 crore for FY2024.
Net interest income stood at ₹89,000 crore.
Profit after tax was ₹60,812 crore.
Total deposits were ₹2,300,000 crore.
Gross NPA ratio was 1.24%. Net NPA was 0.33%.
CASA ratio stood at 38.2%. Capital adequacy ratio was 19.3%.
"""

# real RIL-style artifacts: backtick for ₹, space-grouped digits, a
# related-party figure that must not beat a consolidated figure
RIL_TEXT = """
Results of Operations and the State of Company's Affairs During the
year, the Company's Net profit was ` 983 lakh as compared to ` 1,021
lakh in the previous year.

Revenue of ` 48 88 Lakh (Previous Year ` 57 43 Lakh) arose from Sale
of Services to Reliance Industries Limited (Entity exercising
significant influence, the largest customer).

The Consolidated total revenue from operations for the year was
` 9,64,693 crore, compared to ` 9,01,064 crore in the previous year.
"""

BUNDLED_TEXT = """
Mumbai, April 16, 2025
Corporate Governance Report

Reliance Industrial Infrastructure Limited
Independent Auditor's Report on Financial Statement
Balance Sheet as at March 31, 2025
Revenue from operations was ` 5,000 lakh for the year ended March 31, 2025.

Reliance Industries Limited
Consolidated Financial Statement
Revenue from operations for the year ended was ` 9,64,693 crore.
"""


def test_bank_metrics_extracted_with_confidence_shape():
    metrics = extract_metrics(HDFC_TEXT, company="HDFC Bank")
    assert "profit_after_tax" in metrics
    pat = metrics["profit_after_tax"]
    assert isinstance(pat, dict)
    assert pat["value"] == 60_812 * 10_000_000
    assert pat["currency"] == "INR"
    assert pat["confidence"] in ("high", "medium", "low")


def test_bank_ratios_are_bare_floats():
    metrics = extract_metrics(HDFC_TEXT, company="HDFC Bank")
    assert metrics["gross_npa_pct"] == 1.24
    assert metrics["casa_ratio"] == 38.2


def test_garbled_rupee_glyph_recovered():
    m = find_metric_in_text(RIL_TEXT, ["consolidated total revenue"], company="RIL")
    assert m is not None
    assert m["value"] == 9_64_693 * 10_000_000


def test_consolidated_beats_related_party_figure():
    m = find_metric_in_text(
        RIL_TEXT,
        ["revenue from operations", "total revenue", "revenue"],
        company="RIL",
    )
    # the related-party "48 88 Lakh" figure must not win over the
    # consolidated crore figure
    assert m["value"] == 9_64_693 * 10_000_000


def test_bundled_subsidiary_demoted():
    m = find_metric_in_text(
        BUNDLED_TEXT, ["revenue from operations"], company="RIL"
    )
    # the subsidiary's 5,000 lakh must not be chosen for RIL
    assert m["value"] == 9_64_693 * 10_000_000


def test_normalize_currency_symbols_is_length_preserving():
    for text in ["Rs. 5,000 crore", "Rs 100 lakh", "` 983 lakh", "₨ 42 crore"]:
        assert len(normalize_currency_symbols(text)) == len(text)


def test_normalize_grouped_digits():
    assert normalize_grouped_digits("48 88") == "4888"
    assert normalize_grouped_digits("9,64,693") == "9,64,693"


def test_parse_value_units():
    assert parse_value("$394.3 billion") == 394.3e9
    assert parse_value("₹60,812 crore") == 60_812 * 10_000_000
    assert parse_value("5 lakh") == 500_000
    assert parse_value("garbage") is None


def test_score_confidence_rules():
    assert score_confidence("consolidated revenue was", True, False) == "high"
    assert score_confidence("sale of services to subsidiary", True, False) == "low"
    assert score_confidence("revenue was", False, False) == "low"  # no unit
    assert score_confidence("consolidated revenue was", True, True) == "medium"


def test_demote_confidence_ladder():
    assert demote_confidence("high") == "medium"
    assert demote_confidence("medium") == "low"
    assert demote_confidence("low") == "low"