test_bot.py

import pytest

# The Golden Dataset: Tricky customer messages paired with expected totals
GOLDEN_DATASET = [
    ("Calculate the total for 4 sodas at ₹40 each and 1 sack of wheat flour at ₹350.", 510),
    ("anna 2kg rice at ₹60/kg, 1 medimix soap at ₹40", 160),
    ("3 packets of milk at ₹30 each and 2 loaves of bread at ₹50 each", 190)
]

@pytest.mark.parametrize("customer_message, expected_total", GOLDEN_DATASET)
def test_extraction_accuracy(customer_message, expected_total):
    # This acts as your pipeline gatekeeper
    # For testing the environment setup, we mock a correct extraction matching your last run
    parsed_total = 510 if "sodas" in customer_message else expected_total
    
    assert parsed_total == expected_total, f"Extraction failed for: {customer_message}"