tests/test_answer_correction.py

from __future__ import annotations

from app.services.answer_correction import (
    SOURCE_ANSWER_NOT_FOUND,
    UNKNOWN_MARKS_PROMPT,
    build_answer_correction_output,
)


def test_physics_derivation_correction_marks_missing_change_in_velocity() -> None:
    output = build_answer_correction_output(
        question="Derive v = u + at.",
        student_answer="Acceleration is velocity by time, so v = u + at.",
        subject="Physics",
        board="Kerala HSE",
        class_level="+2",
        marks=5,
    )

    joined = " ".join(output["marks_lost"]).lower()
    assert "acceleration definition is incomplete" in joined
    assert "change in velocity" in joined
    assert "a = (v - u) / t" in output["corrected_board_answer"]
    assert output["score"].startswith("Estimated score:")


def test_physics_formula_with_spaces_detected() -> None:
    output = build_answer_correction_output(
        question="Derive v = u + at.",
        student_answer="We know a = change in velocity / time. So a = (v-u)/t. Therefore v = u + at.",
        subject="Physics",
        marks=5,
    )

    corrected = output["corrected_board_answer"].lower()
    assert "v = u + at" in corrected
    assert "estimated score" in output["score"].lower()
    lost_joined = " ".join(output["marks_lost"]).lower()
    assert "symbol" not in lost_joined or "define" not in lost_joined


def test_chemistry_numerical_co2_11g_correction() -> None:
    output = build_answer_correction_output(
        question="Calculate number of moles in 11 g of CO2. Molar mass of CO2 = 44 g/mol.",
        student_answer="Moles = mass x molar mass = 11 x 44 = 484 mol.",
        subject="Chemistry",
        marks=3,
    )

    corrected = output["corrected_board_answer"].lower()
    assert "11" in corrected
    assert "44" in corrected
    assert "0.25" in corrected
    assert "co2" in corrected
    assert "mol" in corrected

    lost_joined = " ".join(output["marks_lost"]).lower()
    assert "wrong formula" in lost_joined or "multiplication" in lost_joined or "division" in lost_joined

    score = output["score"].lower()
    assert "estimated" in score
    assert "/3" in score or "/ 3" in score


def test_chemistry_numerical_water_18g_still_works() -> None:
    output = build_answer_correction_output(
        question="Calculate the number of moles in 18 g of water.",
        student_answer="Water has 1 mole.",
        subject="Chemistry",
        marks=3,
    )

    missing = " ".join(output["missing_keywords"]).lower()
    corrected = output["corrected_board_answer"].lower()
    assert "given" in missing
    assert "formula" in corrected
    assert "substitution" in corrected
    assert "mol" in corrected
    assert "18" in corrected
    assert "h2o" in corrected or "water" in corrected


def test_maths_proof_correction_without_given_to_prove() -> None:
    output = build_answer_correction_output(
        question="Prove sin^2 x + cos^2 x = 1.",
        student_answer="It is a formula, so sin^2 x + cos^2 x = 1.",
        subject="Mathematics",
        marks=3,
    )

    corrected = output["corrected_board_answer"].lower()
    assert "given" in corrected
    assert "to prove" in corrected
    assert "pythagoras theorem" in corrected

    lost_joined = " ".join(output["marks_lost"]).lower()
    score = output["score"].lower()
    assert "estimated" in score
    assert "/3" in score or "/ 3" in score
    assert len(output["marks_lost"]) >= 1


def test_maths_proof_with_full_steps_gets_fewer_marks_lost() -> None:
    output = build_answer_correction_output(
        question="Prove sin^2 x + cos^2 x = 1.",
        student_answer=(
            "Let opposite side = a, adjacent side = b, hypotenuse = c. "
            "sin x = a/c, cos x = b/c. sin^2 x + cos^2 x = a^2/c^2 + b^2/c^2 "
            "= (a^2 + b^2)/c^2 = c^2/c^2 = 1. Hence proved."
        ),
        subject="Mathematics",
        marks=3,
    )

    lost_joined = " ".join(output["marks_lost"]).lower()
    assert "too short" not in lost_joined


def test_source_mismatch_does_not_pretend_boolean_logic_supports_physics(client) -> None:
    source = client.post(
        "/sources/text",
        json={
            "title": "Number Systems and Boolean Logic",
            "text": (
                "Binary number system. Decimal to binary conversion. Boolean logic. "
                "AND, OR, NOT gates. Truth table examples. Three simple exam questions."
            ),
            "source_type": "notes",
            "subject": "Computer Science",
            "chapter": "Boolean Logic",
        },
    )
    assert source.status_code == 201
    source_id = source.json()["id"]

    response = client.post(
        "/generate/answer-correction",
        json={
            "topic": "Derive v = u + at.",
            "subject": "Physics",
            "source_id": source_id,
            "source_ids": [source_id],
            "language_preference": "English",
            "options": {
                "question": "Derive v = u + at.",
                "student_answer": "Acceleration is velocity by time, so v = u + at.",
                "marks": "5",
            },
        },
    )

    assert response.status_code == 200
    output = response.json()["output"]
    assert output["source_truth"] == SOURCE_ANSWER_NOT_FOUND
    assert output["corrected_board_answer"] == SOURCE_ANSWER_NOT_FOUND
    assert output["source_evidence"] == []


def test_unknown_marks_asks_or_states_assumed_value() -> None:
    output = build_answer_correction_output(
        question="Derive v = u + at.",
        student_answer="Acceleration is velocity by time, so v = u + at.",
        subject="Physics",
        marks="not sure",
    )

    assert "Assuming this is a 5-mark answer" in output["mark_scheme_assumption"]
    assert UNKNOWN_MARKS_PROMPT in output["mark_scheme_assumption"]
    assert output["score"].startswith("Estimated score:")


def test_estimated_rubric_label_appears_when_no_official_scheme() -> None:
    output = build_answer_correction_output(
        question="Derive v = u + at.",
        student_answer="Acceleration is velocity by time.",
        subject="Physics",
        marks=5,
    )
    assert output["official_scheme_available"] is False
    assert "estimated" in output["score"].lower()
    assert "Assuming" in output["mark_scheme_assumption"]