tests/test_math_answer_extractor_scanner.py

from app.schemas.cross_validation import AnswerKind, ValidationConfidence
from app.services.math_answer_extractor import extract_student_math_answer


def test_extractor_uses_scanner_for_set_literal() -> None:
    r = extract_student_math_answer(
        "La reponse est {3}.",
        expected_answer_kind=AnswerKind.SOLUTION_SET,
    )
    assert r.extracted_value == "3"
    assert r.confidence == ValidationConfidence.CERTAIN
    assert r.needs_human_review is False
    assert r.reason == "deterministic_structure_scanner"


def test_extractor_preserves_multi_value_solution_set() -> None:
    r = extract_student_math_answer(
        "On obtient S={-2, 4}.",
        expected_answer_kind=AnswerKind.SOLUTION_SET,
    )
    assert r.extracted_value == "{-2,4}"
    assert r.confidence == ValidationConfidence.CERTAIN
    assert r.needs_human_review is False


def test_extractor_marks_bare_number_as_semantic_residue() -> None:
    r = extract_student_math_answer(
        "Donc 3 est la solution.",
        expected_answer_kind=AnswerKind.SOLUTION_SET,
    )
    assert r.extracted_value == "3"
    assert r.confidence == ValidationConfidence.PROBABLE
    assert r.needs_human_review is True
    assert r.reason == "bare_number_requires_semantic_confirmation"