t2a_v2/tests/test_fusion.py

"""Tests pour le module de fusion multi-PDFs."""

import pytest

from src.config import (
    ActeCCAM,
    Diagnostic,
    DossierMedical,
    Sejour,
    Traitement,
    BiologieCle,
    Imagerie,
)
from src.medical.fusion import (
    merge_dossiers,
    _cim10_specificity,
    _prefer_most_specific_dp,
    _merge_sejour,
    _dedup_diagnostics,
    _dedup_actes,
)


class TestCIM10Specificity:
    def test_none(self):
        assert _cim10_specificity(None) == 0

    def test_short_code(self):
        assert _cim10_specificity("I10") == 3

    def test_long_code(self):
        assert _cim10_specificity("K85.1") == 4

    def test_longer_code(self):
        assert _cim10_specificity("K80.50") == 5


class TestSpecificityLongerCodeWins:
    def test_specificity_longer_code_wins(self):
        d1 = DossierMedical(
            diagnostic_principal=Diagnostic(texte="Calcul biliaire", cim10_suggestion="K80"),
        )
        d2 = DossierMedical(
            diagnostic_principal=Diagnostic(texte="Calcul cholédoque", cim10_suggestion="K80.5"),
        )
        dp = _prefer_most_specific_dp([d1, d2])
        assert dp is not None
        assert dp.cim10_suggestion == "K80.5"


class TestMergeSejourTrackarePriority:
    def test_merge_sejour_trackare_priority(self):
        d1 = DossierMedical(
            document_type="trackare",
            sejour=Sejour(sexe="F", age=43, date_entree="25/02/2023"),
        )
        d2 = DossierMedical(
            document_type="crh",
            sejour=Sejour(sexe="M", age=45, date_entree="24/02/2023", mode_sortie="domicile"),
        )
        merged = _merge_sejour([d1, d2])
        assert merged.sexe == "F"  # Trackare prioritaire
        assert merged.age == 43
        assert merged.date_entree == "25/02/2023"
        assert merged.mode_sortie == "domicile"  # Complété depuis CRH

    def test_merge_sejour_fills_missing(self):
        d1 = DossierMedical(
            document_type="trackare",
            sejour=Sejour(sexe="F"),
        )
        d2 = DossierMedical(
            document_type="crh",
            sejour=Sejour(age=50, poids=75.0),
        )
        merged = _merge_sejour([d1, d2])
        assert merged.sexe == "F"
        assert merged.age == 50
        assert merged.poids == 75.0


class TestDedupDiagnostics:
    def test_dedup_diagnostics_by_code(self):
        das = [
            Diagnostic(texte="HTA", cim10_suggestion="I10", cim10_confidence="medium"),
            Diagnostic(texte="Hypertension", cim10_suggestion="I10", cim10_confidence="high"),
        ]
        result = _dedup_diagnostics(das)
        assert len(result) == 1
        assert result[0].cim10_confidence == "high"

    def test_dedup_keeps_distinct_codes(self):
        das = [
            Diagnostic(texte="HTA", cim10_suggestion="I10"),
            Diagnostic(texte="Diabète", cim10_suggestion="E11.9"),
        ]
        result = _dedup_diagnostics(das)
        assert len(result) == 2


class TestDedupActes:
    def test_dedup_actes_by_code(self):
        actes = [
            ActeCCAM(texte="Cholé", code_ccam_suggestion="HMFC004"),
            ActeCCAM(texte="Cholécystectomie", code_ccam_suggestion="HMFC004", date="01/03"),
        ]
        result = _dedup_actes(actes)
        assert len(result) == 1
        assert result[0].date == "01/03"  # Celui avec la date est préféré


class TestSingleDossierPassthrough:
    def test_single_dossier_passthrough(self):
        d = DossierMedical(
            source_file="test.pdf",
            document_type="crh",
            diagnostic_principal=Diagnostic(texte="HTA", cim10_suggestion="I10"),
        )
        result = merge_dossiers([d])
        assert result.diagnostic_principal.cim10_suggestion == "I10"
        assert result.source_files == ["test.pdf"]


class TestDpNonRetainedBecomesDas:
    def test_dp_non_retained_becomes_das(self):
        d1 = DossierMedical(
            diagnostic_principal=Diagnostic(texte="HTA", cim10_suggestion="I10"),
        )
        d2 = DossierMedical(
            diagnostic_principal=Diagnostic(texte="Calcul cholédoque", cim10_suggestion="K80.5"),
        )
        result = merge_dossiers([d1, d2])
        # K80.5 est plus spécifique, donc DP
        assert result.diagnostic_principal.cim10_suggestion == "K80.5"
        # I10 (ancien DP de d1) doit être dans les DAS
        das_codes = {d.cim10_suggestion for d in result.diagnostics_associes}
        assert "I10" in das_codes


class TestFusionAlertAdded:
    def test_fusion_alert_added(self):
        d1 = DossierMedical(source_file="a.pdf", alertes_codage=["Alerte 1"])
        d2 = DossierMedical(source_file="b.pdf", alertes_codage=["Alerte 2"])
        result = merge_dossiers([d1, d2])
        assert result.alertes_codage[0] == "FUSION: 2 documents fusionnés"
        assert "Alerte 1" in result.alertes_codage
        assert "Alerte 2" in result.alertes_codage


class TestSourceFilesPopulated:
    def test_source_files_populated(self):
        d1 = DossierMedical(source_file="a.pdf")
        d2 = DossierMedical(source_file="b.pdf")
        result = merge_dossiers([d1, d2])
        assert result.source_files == ["a.pdf", "b.pdf"]


class TestDasEqualDpRemoved:
    """Vérifie que les DAS dont le code est identique au DP sont retirés après fusion."""

    def test_das_same_code_as_dp_removed(self):
        d1 = DossierMedical(
            diagnostic_principal=Diagnostic(texte="HTA", cim10_suggestion="I10"),
            diagnostics_associes=[
                Diagnostic(texte="Hypertension artérielle", cim10_suggestion="I10"),
                Diagnostic(texte="Diabète", cim10_suggestion="E11.9"),
            ],
        )
        d2 = DossierMedical(
            diagnostic_principal=Diagnostic(texte="HTA essentielle", cim10_suggestion="I10"),
        )
        result = merge_dossiers([d1, d2])
        das_codes = [d.cim10_suggestion for d in result.diagnostics_associes]
        assert "I10" not in das_codes, "DAS=DP doit être retiré"
        assert "E11.9" in das_codes

    def test_das_different_code_kept(self):
        d1 = DossierMedical(
            diagnostic_principal=Diagnostic(texte="Cholécystite", cim10_suggestion="K81.0"),
            diagnostics_associes=[
                Diagnostic(texte="HTA", cim10_suggestion="I10"),
            ],
        )
        result = merge_dossiers([d1])
        das_codes = [d.cim10_suggestion for d in result.diagnostics_associes]
        assert "I10" in das_codes


class TestFullMergeCROTrackare:
    def test_full_merge_cro_trackare(self):
        """Cas réel : fusion Trackare + CRO."""
        trackare = DossierMedical(
            source_file="trackare.pdf",
            document_type="trackare",
            sejour=Sejour(sexe="F", age=43, date_entree="25/02/2023", date_sortie="03/03/2023"),
            diagnostic_principal=Diagnostic(
                texte="Calcul des canaux biliaires",
                cim10_suggestion="K80.5",
            ),
            diagnostics_associes=[
                Diagnostic(texte="HTA", cim10_suggestion="I10"),
            ],
            actes_ccam=[
                ActeCCAM(texte="Cholécystectomie", code_ccam_suggestion="HMFC004", date="01/03"),
            ],
            traitements_sortie=[
                Traitement(medicament="Paracétamol"),
            ],
            alertes_codage=["Alerte trackare"],
        )

        cro = DossierMedical(
            source_file="cro.pdf",
            document_type="cro",
            sejour=Sejour(sexe="F"),
            diagnostic_principal=Diagnostic(
                texte="Pancréatite aiguë lithiasique",
                cim10_suggestion="K85.1",
                cim10_confidence="high",
            ),
            diagnostics_associes=[
                Diagnostic(texte="Obésité", cim10_suggestion="E66.0"),
                Diagnostic(texte="HTA", cim10_suggestion="I10"),  # doublon
            ],
            actes_ccam=[
                ActeCCAM(texte="TDM", code_ccam_suggestion="ZCQK002"),
                ActeCCAM(texte="Cholécystectomie", code_ccam_suggestion="HMFC004"),  # doublon
            ],
            traitements_sortie=[
                Traitement(medicament="Paracétamol"),  # doublon
                Traitement(medicament="Cétirizine"),
            ],
            alertes_codage=["Alerte CRO"],
        )

        result = merge_dossiers([trackare, cro])

        # DP : K85.1 est plus spécifique que K80.5
        assert result.diagnostic_principal.cim10_suggestion == "K85.1"

        # K80.5 (ancien DP trackare) doit être dans les DAS
        das_codes = {d.cim10_suggestion for d in result.diagnostics_associes}
        assert "K80.5" in das_codes
        assert "I10" in das_codes
        assert "E66.0" in das_codes

        # DAS dédupliqués : I10 ne doit pas être en double
        i10_count = sum(1 for d in result.diagnostics_associes if d.cim10_suggestion == "I10")
        assert i10_count == 1

        # Actes dédupliqués
        acte_codes = [a.code_ccam_suggestion for a in result.actes_ccam]
        assert acte_codes.count("HMFC004") == 1
        assert "ZCQK002" in acte_codes

        # Traitements dédupliqués
        meds = [t.medicament for t in result.traitements_sortie]
        assert meds.count("Paracétamol") == 1
        assert "Cétirizine" in meds

        # Source files
        assert result.source_files == ["trackare.pdf", "cro.pdf"]

        # Alertes
        assert result.alertes_codage[0].startswith("FUSION:")
        assert "Alerte trackare" in result.alertes_codage
        assert "Alerte CRO" in result.alertes_codage

        # Type prioritaire : trackare
        assert result.document_type == "trackare"