feat: cache Ollama + parallélisation ThreadPool + filtrage DAS renforcé + modules GHM/CPAM/export RUM

- Cache persistant JSON thread-safe pour les résultats Ollama (invalidation par modèle)
- Parallélisation des appels Ollama (ThreadPoolExecutor, 2 workers)
- 6 nouvelles règles de filtrage DAS parasites (doublons, ponctuation, OCR, labo, fragments)
- Client Ollama centralisé (mode JSON natif + retry)
- Module GHM (estimation CMD/sévérité)
- Module contrôle CPAM (parser + contre-argumentation RAG)
- Export RUM (format RSS)
- Viewer enrichi (détail dossier)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
dom
2026-02-12 13:44:34 +01:00
parent a00e5f1147
commit a58398f5d4
25 changed files with 2872 additions and 97 deletions

130
tests/test_cpam_parser.py Normal file
View File

@@ -0,0 +1,130 @@
"""Tests pour le parser de contrôle CPAM."""
import tempfile
from pathlib import Path
import openpyxl
import pytest
from src.config import ControleCPAM
from src.control.cpam_parser import match_dossier_ogc, parse_cpam_excel
def _create_test_xlsx(rows: list[tuple], path: Path) -> None:
"""Crée un fichier xlsx de test avec les lignes données."""
wb = openpyxl.Workbook()
ws = wb.active
ws.title = "OGC Contrôle T2A"
ws.append(("N° OGC", "Titre", "Arg_UCR", "Décision_UCR", "DP_UCR", "DA_UCR", "DR_UCR", "Actes_UCR"))
for row in rows:
ws.append(row)
wb.save(path)
class TestParseCpamExcel:
def test_parse_basic(self, tmp_path):
xlsx = tmp_path / "test.xlsx"
_create_test_xlsx([
(17, "Désaccord sur les DAS", "Argument UCR...", "UCR retient", None, None, None, None),
(21, "Désaccord sur le DP", "Autre argument", "UCR confirme avis", "K85.1", None, None, None),
], xlsx)
result = parse_cpam_excel(xlsx)
assert 17 in result
assert 21 in result
assert len(result[17]) == 1
assert len(result[21]) == 1
assert result[17][0].titre == "Désaccord sur les DAS"
assert result[17][0].decision_ucr == "UCR retient"
assert result[21][0].dp_ucr == "K85.1"
def test_parse_multiple_same_ogc(self, tmp_path):
xlsx = tmp_path / "test.xlsx"
_create_test_xlsx([
(17, "Titre 1", "Arg 1", "Décision 1", None, None, None, None),
(17, "Titre 2", "Arg 2", "Décision 2", None, None, None, None),
], xlsx)
result = parse_cpam_excel(xlsx)
assert len(result[17]) == 2
def test_parse_empty_file(self, tmp_path):
xlsx = tmp_path / "empty.xlsx"
_create_test_xlsx([], xlsx)
result = parse_cpam_excel(xlsx)
assert result == {}
def test_parse_nonexistent_file(self):
result = parse_cpam_excel("/nonexistent/path.xlsx")
assert result == {}
def test_parse_optional_fields(self, tmp_path):
xlsx = tmp_path / "test.xlsx"
_create_test_xlsx([
(42, "Titre", "Arg", "Décision", "E11.40", "G63.2", "E11.9", "ABCD123"),
], xlsx)
result = parse_cpam_excel(xlsx)
ctrl = result[42][0]
assert ctrl.dp_ucr == "E11.40"
assert ctrl.da_ucr == "G63.2"
assert ctrl.dr_ucr == "E11.9"
assert ctrl.actes_ucr == "ABCD123"
class TestMatchDossierOGC:
def setup_method(self):
self.cpam_data = {
17: [ControleCPAM(numero_ogc=17, titre="Test 17")],
21: [ControleCPAM(numero_ogc=21, titre="Test 21")],
}
def test_match_found(self):
result = match_dossier_ogc("17_23100690", self.cpam_data)
assert len(result) == 1
assert result[0].numero_ogc == 17
def test_match_not_found(self):
result = match_dossier_ogc("15_23096332", self.cpam_data)
assert result == []
def test_match_no_prefix(self):
result = match_dossier_ogc("nodash", self.cpam_data)
assert result == []
def test_match_empty_data(self):
result = match_dossier_ogc("17_23100690", {})
assert result == []
class TestControleCPAMModel:
def test_serialization(self):
ctrl = ControleCPAM(
numero_ogc=17,
titre="Désaccord sur les DAS",
arg_ucr="Argument...",
decision_ucr="UCR retient",
dp_ucr="K85.1",
)
data = ctrl.model_dump()
assert data["numero_ogc"] == 17
assert data["dp_ucr"] == "K85.1"
assert data["contre_argumentation"] is None
def test_deserialization(self):
data = {
"numero_ogc": 21,
"titre": "Test",
"arg_ucr": "Arg",
"decision_ucr": "Décision",
"contre_argumentation": "Ma réponse",
}
ctrl = ControleCPAM(**data)
assert ctrl.numero_ogc == 21
assert ctrl.contre_argumentation == "Ma réponse"
assert ctrl.sources_reponse == []