Aivanov_scan_ogc/tests/test_validation.py

"""Tests unitaires pour pipeline.validation."""
from __future__ import annotations

import pytest

from pipeline.validation import (
    _check_ccam,
    _check_cim10,
    _check_ghm,
    _check_ghs,
    _cross_check_ghm_ghs,
    annotate,
    validate_recueil,
)


# ============================================================
# Vérifications par type de code
# ============================================================

class TestCheckCim10:
    def test_code_valide(self):
        r = _check_cim10("K650")
        assert r["valid"] is True
        assert "libelle_ref" in r

    def test_code_vide(self):
        assert _check_cim10("")["valid"] is None
        assert _check_cim10(None)["valid"] is None

    def test_code_avec_suffixe_pmsi(self):
        # Les suffixes * et +N sont gérés par la normalisation
        r = _check_cim10("C795 *")
        assert r["valid"] is True

    def test_code_invalide_avec_suggestion(self):
        # K65O (O au lieu de 0) n'existe pas, mais K650 oui
        r = _check_cim10("K65O")
        assert r["valid"] is False
        assert r.get("suggestion") == "K650"

    def test_code_invalide_sans_suggestion(self):
        # Code farfelu sans voisin proche
        r = _check_cim10("ZZZZ9999")
        assert r["valid"] is False
        # suggestion peut être absente
        assert r.get("suggestion") is None or r.get("suggestion") != "ZZZZ9999"


class TestCheckGhm:
    def test_ghm_valide(self):
        r = _check_ghm("11M122")
        assert r["valid"] is True
        assert isinstance(r.get("ghs_possibles"), list)
        assert len(r["ghs_possibles"]) > 0

    def test_ghm_invalide(self):
        r = _check_ghm("99Z999")
        assert r["valid"] is False


class TestCheckGhs:
    def test_ghs_valide(self):
        assert _check_ghs("4323")["valid"] is True

    def test_ghs_invalide(self):
        assert _check_ghs("99999")["valid"] is False


class TestCheckCcam:
    def test_ccam_valide(self):
        assert _check_ccam("EBFA012")["valid"] is True

    def test_ccam_invalide(self):
        assert _check_ccam("XXXX000")["valid"] is False


# ============================================================
# Cross-checks GHM ↔ GHS
# ============================================================

class TestCrossCheckGhmGhs:
    def test_couple_coherent(self):
        # 11M122 a bien 4323 dans ses GHS possibles
        r = _cross_check_ghm_ghs("11M122", "4323")
        assert r["checked"] is True
        assert r["coherent"] is True

    def test_couple_incoherent(self):
        # 11M122 ne correspond pas à n'importe quel GHS
        r = _cross_check_ghm_ghs("11M122", "9999")
        assert r["checked"] is True
        assert r["coherent"] is False

    def test_ghm_manquant(self):
        r = _cross_check_ghm_ghs("", "4323")
        assert r["checked"] is False

    def test_ghm_invalide(self):
        r = _cross_check_ghm_ghs("99Z999", "4323")
        assert r["checked"] is False
        assert "invalide" in r["reason"].lower()


# ============================================================
# annotate (intégration)
# ============================================================

class TestAnnotate:
    def test_annotate_json_vide(self):
        out = annotate({"fichier": "TEST", "extraction": {}})
        assert "fichier" in out
        assert out["extraction"] == {}

    def test_annotate_recueil_complet(self):
        raw = {
            "fichier": "TEST",
            "extraction": {
                "recueil": {
                    "codage_etab": {"dp": "K650", "dr": "", "das": [
                        {"code": "T814", "position": "2"},
                    ]},
                    "codage_reco": {"dp": "", "dr": "", "das": []},
                    "ghm_etab": "11M122",
                    "ghs_etab": "4323",
                    "ghm_reco": "",
                    "ghs_reco": "",
                },
            },
        }
        out = annotate(raw)
        v = out["extraction"]["recueil"]["_validation"]
        assert v["codage_etab"]["dp"]["valid"] is True
        assert v["ghm_etab"]["valid"] is True
        assert v["cross_checks"]["etab"]["coherent"] is True
        assert v["summary"]["valid"] >= 3

    def test_annotate_preserve_source(self):
        """L'annotation ne doit pas modifier l'input (copie défensive)."""
        raw = {
            "fichier": "T",
            "extraction": {"recueil": {"codage_etab": {"dp": "K650"}}},
        }
        out = annotate(raw)
        assert "_validation" not in raw["extraction"]["recueil"]
        assert "_validation" in out["extraction"]["recueil"]