app/services/classifier.py

from app.core.taxonomy import AnswerType, Chapter, Level, Subject
from app.schemas.classification import ClassificationResult


CHAPTER_KEYWORDS: tuple[tuple[Chapter, tuple[str, ...]], ...] = (
    (
        Chapter.DERIVATION,
        (
            "dérivée",
            "derivee",
            "dérivable",
            "derivable",
            "dériver",
            "derive",
            "f'(x)",
            "f’(x)",
            "signe de la dérivée",
            "signe de f'",
            "tableau de variation",
        ),
    ),
    (
        Chapter.LIMITES,
        (
            "limite",
            "tend vers",
            "infini",
            "+∞",
            "-∞",
            "asymptote",
        ),
    ),
    (
        Chapter.SUITES,
        (
            "suite",
            "u_n",
            "un+1",
            "récurrence",
            "recurrence",
            "arithmétique",
            "géométrique",
        ),
    ),
    (
        Chapter.FONCTIONS,
        (
            "fonction",
            "image",
            "antécédent",
            "ensemble de définition",
            "courbe",
        ),
    ),
)


def detect_answer_type(answer_text: str) -> AnswerType:
    if "\\" in answer_text or "$" in answer_text:
        return AnswerType.LATEX_SIMPLE
    if any(marker in answer_text for marker in ("#", "-", "*", "`")):
        return AnswerType.MARKDOWN
    return AnswerType.TEXTE


def classify_submission_answer(answer_text: str) -> ClassificationResult:
    normalized = answer_text.lower()
    detected_chapter = Chapter.FONCTIONS
    confidence = 0.55
    skill = "raisonner sur une fonction"

    for chapter, keywords in CHAPTER_KEYWORDS:
        if any(keyword in normalized for keyword in keywords):
            detected_chapter = chapter
            confidence = 0.82
            if chapter == Chapter.DERIVATION:
                skill = "calculer une dérivée et exploiter son signe"
            elif chapter == Chapter.LIMITES:
                skill = "calculer une limite"
            elif chapter == Chapter.SUITES:
                skill = "étudier une suite numérique"
            else:
                skill = "étudier une fonction"
            break

    return ClassificationResult(
        subject=Subject.MATHEMATIQUES,
        level=Level.DEUXIEME_BAC_SCIENCES,
        chapter=detected_chapter,
        skill=skill,
        answer_type=detect_answer_type(answer_text),
        language="fr",
        confidence=confidence,
        reject_reason=None,
    )