feat: enrichissement contre-argumentation CPAM — libellés CIM-10, RAG ciblé, reprocess complet

- Résolution des libellés CIM-10 pour les codes contestés (dp_ucr, da_ucr, dr_ucr)
- Fallback DP depuis dp_ucr quand le pipeline n'extrait pas de diagnostic principal
- Troncature arg_ucr augmentée de 200 à 500 chars pour conserver les citations de règles
- Requête RAG 4 : définitions CIM-10 (inclusion/exclusion) des codes contestés
- Requête RAG 5 : extraction et recherche des règles nommées (RègleT7, Annexe, etc.)
- Cap résultats RAG de 10 à 12 pour absorber les nouvelles requêtes
- Reprocess viewer : pipeline complet (fusion + GHM + CPAM) pour dossiers multi-PDF
- Affichage structuré response_data dans le viewer (analyse, preuves, références)
- 7 nouveaux tests CPAM, 6 nouveaux tests viewer

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
dom
2026-02-17 23:24:10 +01:00
parent 94fa4e5f3b
commit bc0ccbef7c
7 changed files with 464 additions and 51 deletions

View File

@@ -188,6 +188,7 @@ class ControleCPAM(BaseModel):
dr_ucr: Optional[str] = None
actes_ucr: Optional[str] = None
contre_argumentation: Optional[str] = None
response_data: Optional[dict] = None
sources_reponse: list[RAGSource] = Field(default_factory=list)

View File

@@ -3,8 +3,10 @@
from __future__ import annotations
import logging
import re
from ..config import ControleCPAM, DossierMedical, RAGSource
from ..medical.cim10_dict import normalize_code, validate_code
from ..medical.ollama_client import call_anthropic, call_ollama
logger = logging.getLogger(__name__)
@@ -42,7 +44,7 @@ def _search_rag_for_control(controle: ControleCPAM, dossier: DossierMedical) ->
query_parts_arg = []
if controle.titre:
query_parts_arg.append(controle.titre)
arg_short = controle.arg_ucr[:200] if controle.arg_ucr else ""
arg_short = controle.arg_ucr[:500] if controle.arg_ucr else ""
if arg_short:
query_parts_arg.append(arg_short)
query_arg = " ".join(query_parts_arg)
@@ -65,6 +67,43 @@ def _search_rag_for_control(controle: ControleCPAM, dossier: DossierMedical) ->
logger.debug(" RAG requête clinique : %d résultats", len(results_clinique))
all_results.extend(results_clinique)
# Requête 4 — Définitions CIM-10 des codes contestés
contested_codes = []
for field in (controle.dp_ucr, controle.da_ucr, controle.dr_ucr):
if field:
contested_codes.extend(re.split(r"[,;\s]+", field.strip()))
for raw_code in contested_codes:
raw_code = raw_code.strip()
if not raw_code:
continue
norm = normalize_code(raw_code)
is_valid, label = validate_code(norm)
if is_valid and label:
query_def = f"CIM-10 {norm} {label} définition inclusion exclusion"
else:
query_def = f"CIM-10 {norm} définition codage"
results_def = search_similar_cpam(query_def, top_k=3)
logger.debug(" RAG requête CIM-10 %s : %d résultats", norm, len(results_def))
all_results.extend(results_def)
# Requête 5 — Règles explicitement citées dans l'argument CPAM
if controle.arg_ucr:
rule_patterns = [
r'(?:R[eè]gle\s*T?\s*\d+)',
r'(?:Annexe[\s-]*\d+[A-Za-z]*)',
r'(?:Situation de soins?\s+[^.]{5,40})',
]
rules_found = []
for pattern in rule_patterns:
rules_found.extend(re.findall(pattern, controle.arg_ucr, re.IGNORECASE))
if rules_found:
rules_unique = list(dict.fromkeys(rules_found))
query_rules = " ".join(rules_unique) + " guide méthodologique codage PMSI"
results_rules = search_similar_cpam(query_rules, top_k=4)
logger.debug(" RAG requête règles (%s) : %d résultats",
", ".join(rules_unique), len(results_rules))
all_results.extend(results_rules)
if not all_results:
return []
@@ -79,7 +118,29 @@ def _search_rag_for_control(controle: ControleCPAM, dossier: DossierMedical) ->
seen[key] = r
merged = sorted(seen.values(), key=lambda r: r["score"], reverse=True)
return merged[:10]
return merged[:12]
def _get_code_label(code_str: str) -> str:
"""Résout le libellé CIM-10 pour un ou plusieurs codes."""
codes = re.split(r"[,;\s]+", code_str.strip())
labels = []
for raw in codes:
raw = raw.strip()
if not raw:
continue
norm = normalize_code(raw)
is_valid, label = validate_code(norm)
if is_valid and label:
labels.append(f"{norm}{label}")
else:
labels.append(norm)
if not labels:
return ""
if len(labels) == 1:
parts = labels[0].split("", 1)
return f"{parts[1]}" if len(parts) > 1 else ""
return "\n " + "\n ".join(labels)
def _build_cpam_prompt(
@@ -95,6 +156,12 @@ def _build_cpam_prompt(
dp = dossier.diagnostic_principal
dp_code = f" ({dp.cim10_suggestion})" if dp.cim10_suggestion else ""
dossier_lines.append(f"- DP : {dp.texte}{dp_code}")
elif controle.dp_ucr:
dp_label = _get_code_label(controle.dp_ucr)
dossier_lines.append(
f"- DP : code {controle.dp_ucr}{dp_label} "
f"(codé par l'établissement, contesté par la CPAM)"
)
if dossier.diagnostics_associes:
das_parts = []
@@ -192,14 +259,14 @@ def _build_cpam_prompt(
+ "\n".join(asymetrie_lines)
)
# Codes contestés par la CPAM
# Codes contestés par la CPAM (avec libellés CIM-10 résolus)
codes_contestes = []
if controle.dp_ucr:
codes_contestes.append(f"DP proposé par UCR : {controle.dp_ucr}")
codes_contestes.append(f"DP proposé par UCR : {controle.dp_ucr}{_get_code_label(controle.dp_ucr)}")
if controle.da_ucr:
codes_contestes.append(f"DA proposés par UCR : {controle.da_ucr}")
codes_contestes.append(f"DA proposés par UCR : {controle.da_ucr}{_get_code_label(controle.da_ucr)}")
if controle.dr_ucr:
codes_contestes.append(f"DR proposé par UCR : {controle.dr_ucr}")
codes_contestes.append(f"DR proposé par UCR : {controle.dr_ucr}{_get_code_label(controle.dr_ucr)}")
if controle.actes_ucr:
codes_contestes.append(f"Actes proposés par UCR : {controle.actes_ucr}")
codes_str = "\n".join(codes_contestes) if codes_contestes else "Aucun code spécifique proposé"
@@ -263,6 +330,10 @@ AXE ASYMÉTRIE D'INFORMATION :
- Démontre en quoi ces éléments complémentaires (biologie, imagerie, traitements, actes) justifient le codage contesté
- Ne mentionne AUCUN élément qui n'est pas dans le dossier fourni
MISE EN FORME :
- Structure chaque section avec des tirets pour lister les arguments distincts
- Un argument par puce, avec la preuve ou la référence associée
AXE RÉGLEMENTAIRE :
- Identifie si l'UCR fait une interprétation restrictive non fondée d'une règle
- Confronte le raisonnement CPAM au texte EXACT des sources fournies
@@ -406,7 +477,7 @@ def _format_response(parsed: dict, ref_warnings: list[str] | None = None) -> str
def generate_cpam_response(
dossier: DossierMedical,
controle: ControleCPAM,
) -> tuple[str, list[RAGSource]]:
) -> tuple[str, dict | None, list[RAGSource]]:
"""Génère une contre-argumentation pour un contrôle CPAM.
Args:
@@ -414,7 +485,7 @@ def generate_cpam_response(
controle: Le contrôle CPAM à contester.
Returns:
Tuple (texte de contre-argumentation, sources RAG utilisées).
Tuple (texte de contre-argumentation, dict LLM structuré ou None, sources RAG utilisées).
"""
logger.info("CPAM : génération contre-argumentation pour OGC %d%s",
controle.numero_ogc, controle.titre)
@@ -449,7 +520,7 @@ def generate_cpam_response(
if result is None:
logger.warning(" LLM non disponible — contre-argumentation non générée")
return "", rag_sources
return "", None, rag_sources
# 5. Validation des références
ref_warnings = _validate_references(result, sources)
@@ -460,4 +531,4 @@ def generate_cpam_response(
text = _format_response(result, ref_warnings)
logger.info(" Contre-argumentation générée (%d caractères)", len(text))
return text, rag_sources
return text, result, rag_sources

View File

@@ -378,8 +378,9 @@ def main(input_path: str | None = None) -> None:
if target:
logger.info(" CPAM : %d contrôle(s) pour %s", len(controles), subdir)
for ctrl in controles:
text, sources = generate_cpam_response(target, ctrl)
text, response_data, sources = generate_cpam_response(target, ctrl)
ctrl.contre_argumentation = text
ctrl.response_data = response_data
ctrl.sources_reponse = sources
target.controles_cpam = controles
except Exception:

View File

@@ -373,6 +373,37 @@ def format_doc_name(name: str) -> str:
return name
def format_cpam_text(text: str | None) -> Markup:
"""Convertit un texte CPAM (section) en HTML avec puces et paragraphes."""
if not text:
return Markup("")
from markupsafe import escape
lines = str(text).split("\n")
html_parts: list[str] = []
in_list = False
for line in lines:
stripped = line.strip()
if not stripped:
if in_list:
html_parts.append("</ul>")
in_list = False
html_parts.append("<br>")
continue
if stripped.startswith("- "):
if not in_list:
html_parts.append("<ul style='margin:0.3rem 0;padding-left:1.2rem;'>")
in_list = True
html_parts.append(f"<li>{escape(stripped[2:])}</li>")
else:
if in_list:
html_parts.append("</ul>")
in_list = False
html_parts.append(f"<p style='margin:0.2rem 0;'>{escape(stripped)}</p>")
if in_list:
html_parts.append("</ul>")
return Markup("\n".join(html_parts))
# ---------------------------------------------------------------------------
# App factory
# ---------------------------------------------------------------------------
@@ -387,6 +418,7 @@ def create_app() -> Flask:
app.jinja_env.filters["format_duration"] = format_duration
app.jinja_env.filters["format_dossier_name"] = format_dossier_name
app.jinja_env.filters["format_doc_name"] = format_doc_name
app.jinja_env.filters["format_cpam_text"] = format_cpam_text
ccam_dict = load_ccam_dict()
@@ -445,36 +477,119 @@ def create_app() -> Flask:
@app.route("/reprocess/<path:filepath>", methods=["POST"])
def reprocess(filepath: str):
"""Relance le traitement d'un dossier."""
"""Relance le pipeline complet : process PDFs + fusion + GHM + CPAM."""
from ..main import process_pdf, write_outputs
from ..medical.ghm import estimate_ghm
dossier = load_dossier(filepath)
source_file = dossier.source_file
if not source_file:
input_dir = Path(__file__).parent.parent.parent / "input"
# Collecter les PDFs sources (fusionné → source_files, simple → source_file)
source_names = []
if dossier.source_files:
source_names = list(dossier.source_files)
elif dossier.source_file:
source_names = [dossier.source_file]
if not source_names:
return jsonify({"error": "Fichier source introuvable"}), 400
# Chercher le PDF source dans input/
input_dir = Path(__file__).parent.parent.parent / "input"
pdf_path = None
for p in input_dir.rglob(source_file):
if p.is_file():
pdf_path = p
break
# Résoudre les chemins PDF dans input/
pdf_paths = []
missing = []
for name in source_names:
found = None
for p in input_dir.rglob(name):
if p.is_file():
found = p
break
if found:
pdf_paths.append(found)
else:
missing.append(name)
if not pdf_path:
return jsonify({"error": f"PDF source '{source_file}' introuvable"}), 404
if not pdf_paths:
return jsonify({"error": f"PDF sources introuvables : {', '.join(missing)}"}), 404
try:
pdf_results = process_pdf(pdf_path)
stem = pdf_path.stem.replace(" ", "_")
# Déterminer le subdir depuis le premier PDF trouvé
subdir = None
if pdf_path.parent != input_dir:
subdir = pdf_path.parent.name
multi = len(pdf_results) > 1
for part_idx, (anonymized_text, new_dossier, report) in enumerate(pdf_results):
part_stem = f"{stem}_part{part_idx + 1}" if multi else stem
write_outputs(part_stem, anonymized_text, new_dossier, report, subdir=subdir)
return jsonify({"ok": True, "message": f"Traitement terminé ({len(pdf_results)} dossier(s))"})
if pdf_paths[0].parent != input_dir:
subdir = pdf_paths[0].parent.name
# 1. Traiter chaque PDF
group_dossiers = []
for pdf_path in pdf_paths:
pdf_results = process_pdf(pdf_path)
stem = pdf_path.stem.replace(" ", "_")
multi = len(pdf_results) > 1
for part_idx, (anonymized_text, new_dossier, report) in enumerate(pdf_results):
part_stem = f"{stem}_part{part_idx + 1}" if multi else stem
write_outputs(part_stem, anonymized_text, new_dossier, report, subdir=subdir)
group_dossiers.append(new_dossier)
# 2. Fusion multi-PDF
merged = None
if len(group_dossiers) > 1 and subdir:
try:
from ..medical.fusion import merge_dossiers
merged = merge_dossiers(group_dossiers)
try:
ghm = estimate_ghm(merged)
merged.ghm_estimation = ghm
except Exception:
logger.warning("Erreur estimation GHM fusionné", exc_info=True)
except Exception:
logger.exception("Erreur fusion groupe %s", subdir)
# 3. Contrôle CPAM (auto-détection Excel)
target = merged if merged else (group_dossiers[-1] if group_dossiers else None)
if target and subdir:
cpam_dir = input_dir / "Control_cpam"
cpam_path = None
if cpam_dir.is_dir():
xlsx_files = sorted(cpam_dir.glob("*.xlsx"))
if xlsx_files:
cpam_path = xlsx_files[0]
if cpam_path:
try:
from ..control.cpam_parser import parse_cpam_excel, match_dossier_ogc
from ..control.cpam_response import generate_cpam_response
cpam_data = parse_cpam_excel(str(cpam_path))
if cpam_data:
controles = match_dossier_ogc(subdir, cpam_data)
if controles:
logger.info("CPAM reprocess : %d contrôle(s) pour %s",
len(controles), subdir)
for ctrl in controles:
text, response_data, sources = generate_cpam_response(target, ctrl)
ctrl.contre_argumentation = text
ctrl.response_data = response_data
ctrl.sources_reponse = sources
target.controles_cpam = controles
except Exception:
logger.exception("Erreur CPAM reprocess pour %s", subdir)
# 4. Écrire le dossier fusionné (après CPAM)
if merged is not None and subdir:
struct_dir = STRUCTURED_DIR / subdir
struct_dir.mkdir(parents=True, exist_ok=True)
merged_path = struct_dir / f"{subdir}_fusionne_cim10.json"
merged_path.write_text(
merged.model_dump_json(indent=2, exclude_none=True),
encoding="utf-8",
)
logger.info("Dossier fusionné réécrit : %s", merged_path)
msg = f"Traitement terminé ({len(group_dossiers)} dossier(s)"
if merged:
msg += ", fusionné"
if target and getattr(target, "controles_cpam", None):
msg += f", {len(target.controles_cpam)} contrôle(s) CPAM"
if missing:
msg += f", {len(missing)} PDF(s) manquant(s)"
msg += ")"
return jsonify({"ok": True, "message": msg})
except Exception as e:
logger.exception("Erreur lors du retraitement")
return jsonify({"error": str(e)}), 500

View File

@@ -165,8 +165,86 @@
</div>
{% endif %}
{# Contre-argumentation #}
{% if ctrl.contre_argumentation %}
{# Contre-argumentation structurée ou fallback texte brut #}
{% if ctrl.response_data %}
<div style="margin-bottom:0.75rem;">
<div style="font-size:0.7rem;color:#1d4ed8;text-transform:uppercase;font-weight:600;margin-bottom:0.5rem;">Contre-argumentation</div>
{% if ctrl.response_data.analyse_contestation %}
<div style="border-left:3px solid #94a3b8;padding:0.5rem 0.75rem;background:#f8fafc;margin-bottom:0.5rem;font-size:0.85rem;">
<div style="font-size:0.7rem;color:#64748b;text-transform:uppercase;font-weight:600;margin-bottom:0.25rem;">Analyse de la contestation</div>
{{ ctrl.response_data.analyse_contestation | format_cpam_text }}
</div>
{% endif %}
{% if ctrl.response_data.points_accord and ctrl.response_data.points_accord|lower not in ['aucun', 'non applicable', 'n/a', ''] %}
<div style="border-left:3px solid #22c55e;padding:0.5rem 0.75rem;background:#f0fdf4;margin-bottom:0.5rem;font-size:0.85rem;">
<div style="font-size:0.7rem;color:#16a34a;text-transform:uppercase;font-weight:600;margin-bottom:0.25rem;">Points d'accord</div>
{{ ctrl.response_data.points_accord | format_cpam_text }}
</div>
{% endif %}
{% if ctrl.response_data.contre_arguments_medicaux %}
<div style="border-left:3px solid #3b82f6;padding:0.5rem 0.75rem;background:#eff6ff;margin-bottom:0.5rem;font-size:0.85rem;">
<div style="font-size:0.7rem;color:#1d4ed8;text-transform:uppercase;font-weight:600;margin-bottom:0.25rem;">Contre-arguments médicaux</div>
{{ ctrl.response_data.contre_arguments_medicaux | format_cpam_text }}
</div>
{% endif %}
{% if ctrl.response_data.preuves_dossier %}
<div style="border-left:3px solid #0ea5e9;padding:0.5rem 0.75rem;background:#f0f9ff;margin-bottom:0.5rem;font-size:0.85rem;">
<div style="font-size:0.7rem;color:#0369a1;text-transform:uppercase;font-weight:600;margin-bottom:0.25rem;">Preuves du dossier</div>
<ul style="margin:0.3rem 0;padding-left:1.2rem;">
{% for p in ctrl.response_data.preuves_dossier %}
{% if p is mapping %}
<li style="margin-bottom:0.3rem;">
<span style="display:inline-block;padding:1px 6px;border-radius:9999px;font-size:0.7rem;font-weight:600;background:#e0f2fe;color:#0369a1;">{{ p.element or p.get('type', '') }}</span>
{{ p.valeur or '' }} <span style="color:#64748b;">&rarr; {{ p.signification or '' }}</span>
</li>
{% endif %}
{% endfor %}
</ul>
</div>
{% endif %}
{% if ctrl.response_data.contre_arguments_asymetrie %}
<div style="border-left:3px solid #8b5cf6;padding:0.5rem 0.75rem;background:#f5f3ff;margin-bottom:0.5rem;font-size:0.85rem;">
<div style="font-size:0.7rem;color:#7c3aed;text-transform:uppercase;font-weight:600;margin-bottom:0.25rem;">Asymétrie d'information</div>
{{ ctrl.response_data.contre_arguments_asymetrie | format_cpam_text }}
</div>
{% endif %}
{% if ctrl.response_data.contre_arguments_reglementaires %}
<div style="border-left:3px solid #6366f1;padding:0.5rem 0.75rem;background:#eef2ff;margin-bottom:0.5rem;font-size:0.85rem;">
<div style="font-size:0.7rem;color:#4f46e5;text-transform:uppercase;font-weight:600;margin-bottom:0.25rem;">Contre-arguments réglementaires</div>
{{ ctrl.response_data.contre_arguments_reglementaires | format_cpam_text }}
</div>
{% endif %}
{% if ctrl.response_data.references %}
<div style="border-left:3px solid #64748b;padding:0.5rem 0.75rem;background:#f8fafc;margin-bottom:0.5rem;font-size:0.85rem;">
<div style="font-size:0.7rem;color:#475569;text-transform:uppercase;font-weight:600;margin-bottom:0.25rem;">Références</div>
{% for ref in ctrl.response_data.references %}
{% if ref is mapping %}
<blockquote style="margin:0.3rem 0;padding:0.3rem 0.5rem;border-left:2px solid #cbd5e1;background:#f1f5f9;font-size:0.8rem;color:#334155;">
<strong>[{{ ref.document or '' }}{% if ref.page %}, p.{{ ref.page }}{% endif %}]</strong>
{{ ref.citation or '' }}
</blockquote>
{% elif ref is string %}
<p style="margin:0.2rem 0;font-size:0.8rem;color:#334155;">{{ ref }}</p>
{% endif %}
{% endfor %}
</div>
{% endif %}
{% if ctrl.response_data.conclusion %}
<div style="border-left:3px solid #f59e0b;padding:0.5rem 0.75rem;background:#fffbeb;margin-bottom:0.5rem;font-size:0.85rem;border:1px solid #fde68a;border-left:3px solid #f59e0b;border-radius:0.25rem;">
<div style="font-size:0.7rem;color:#b45309;text-transform:uppercase;font-weight:600;margin-bottom:0.25rem;">Conclusion</div>
{{ ctrl.response_data.conclusion | format_cpam_text }}
</div>
{% endif %}
</div>
{% elif ctrl.contre_argumentation %}
<div style="border-left:3px solid #3b82f6;padding:0.5rem 0.75rem;background:#eff6ff;margin-bottom:0.75rem;font-size:0.85rem;color:#1e3a5f;">
<div style="font-size:0.7rem;color:#1d4ed8;text-transform:uppercase;font-weight:600;margin-bottom:0.25rem;">Contre-argumentation</div>
<pre style="white-space:pre-wrap;font-family:inherit;margin:0;">{{ ctrl.contre_argumentation }}</pre>

View File

@@ -18,6 +18,7 @@ from src.config import (
from src.control.cpam_response import (
_build_cpam_prompt,
_format_response,
_get_code_label,
_search_rag_for_control,
_validate_references,
generate_cpam_response,
@@ -199,6 +200,51 @@ class TestBuildPrompt:
assert "preuves_dossier" in prompt
@patch("src.control.cpam_response.validate_code", return_value=(True, "Iléus paralytique et obstruction intestinale"))
@patch("src.control.cpam_response.normalize_code", return_value="K56.0")
def test_prompt_codes_with_cim10_labels(self, mock_norm, mock_valid):
"""Les codes contestés affichent le libellé CIM-10."""
dossier = _make_dossier()
controle = _make_controle() # da_ucr="K56.0"
prompt = _build_cpam_prompt(dossier, controle, [])
assert "Iléus paralytique" in prompt
assert "DA proposés par UCR" in prompt
@patch("src.control.cpam_response.validate_code", return_value=(False, ""))
@patch("src.control.cpam_response.normalize_code", return_value="Z99.9")
def test_prompt_codes_invalid_graceful(self, mock_norm, mock_valid):
"""Les codes invalides ne crashent pas, juste pas de libellé."""
dossier = _make_dossier()
controle = ControleCPAM(
numero_ogc=1, titre="Test", arg_ucr="Test",
decision_ucr="Rejet", dp_ucr="Z99.9", da_ucr=None,
)
prompt = _build_cpam_prompt(dossier, controle, [])
assert "Z99.9" in prompt
# Pas de crash
@patch("src.control.cpam_response.validate_code", return_value=(True, "Ajustement et entretien d'un dispositif implantable"))
@patch("src.control.cpam_response.normalize_code", return_value="Z45.8")
def test_prompt_dp_fallback_from_ucr(self, mock_norm, mock_valid):
"""DP absent + dp_ucr → contexte injecté dans le prompt."""
dossier = DossierMedical(
source_file="test.pdf",
document_type="crh",
sejour=Sejour(),
diagnostic_principal=None,
)
controle = ControleCPAM(
numero_ogc=1, titre="Désaccord DP", arg_ucr="Test",
decision_ucr="Rejet", dp_ucr="Z45.8", da_ucr=None,
)
prompt = _build_cpam_prompt(dossier, controle, [])
assert "codé par l'établissement" in prompt
assert "contesté par la CPAM" in prompt
assert "Z45.8" in prompt
class TestFormatResponse:
def test_full_response_new_format(self):
@@ -349,7 +395,7 @@ class TestGenerateResponse:
@patch("src.control.cpam_response.call_anthropic")
@patch("src.control.cpam_response._search_rag_for_control")
def test_generate_success_ollama_cpam(self, mock_rag, mock_anthropic, mock_ollama):
"""Mode hybride : Ollama CPAM (27b) disponible → utilisé en premier."""
"""Ollama disponible → utilisé en premier, retourne triplet."""
mock_rag.return_value = [
{"document": "guide_methodo", "page": 64, "extrait": "Texte guide"},
]
@@ -363,12 +409,14 @@ class TestGenerateResponse:
dossier = _make_dossier()
controle = _make_controle()
text, sources = generate_cpam_response(dossier, controle)
text, response_data, sources = generate_cpam_response(dossier, controle)
assert "Contre-arguments médicaux..." in text
assert response_data is not None
assert response_data["analyse_contestation"] == "Analyse..."
assert response_data["conclusion"] == "Conclusion..."
assert len(sources) == 1
assert sources[0].document == "guide_methodo"
# Ollama CPAM appelé en premier (avec model= et timeout=)
mock_ollama.assert_called_once()
mock_anthropic.assert_not_called()
@@ -376,7 +424,7 @@ class TestGenerateResponse:
@patch("src.control.cpam_response.call_anthropic")
@patch("src.control.cpam_response._search_rag_for_control")
def test_generate_fallback_haiku(self, mock_rag, mock_anthropic, mock_ollama):
"""Ollama CPAM indisponible → fallback Haiku."""
"""Ollama indisponible → fallback Haiku, retourne triplet."""
mock_rag.return_value = [
{"document": "guide_methodo", "page": 64, "extrait": "Texte guide"},
]
@@ -390,10 +438,11 @@ class TestGenerateResponse:
dossier = _make_dossier()
controle = _make_controle()
text, sources = generate_cpam_response(dossier, controle)
text, response_data, sources = generate_cpam_response(dossier, controle)
assert "Contre-args Haiku..." in text
# Ollama CPAM appelé d'abord (échec), puis Haiku
assert response_data is not None
assert response_data["contre_arguments_medicaux"] == "Contre-args Haiku..."
mock_ollama.assert_called_once()
mock_anthropic.assert_called_once()
@@ -401,7 +450,7 @@ class TestGenerateResponse:
@patch("src.control.cpam_response.call_anthropic")
@patch("src.control.cpam_response._search_rag_for_control")
def test_generate_all_unavailable(self, mock_rag, mock_anthropic, mock_ollama):
"""Ollama CPAM, Haiku et Ollama défaut tous indisponibles → texte vide."""
"""Tous LLMs indisponibles → texte vide, response_data None."""
mock_rag.return_value = []
mock_anthropic.return_value = None
mock_ollama.return_value = None
@@ -409,9 +458,10 @@ class TestGenerateResponse:
dossier = _make_dossier()
controle = _make_controle()
text, sources = generate_cpam_response(dossier, controle)
text, response_data, sources = generate_cpam_response(dossier, controle)
assert text == ""
assert response_data is None
assert sources == []
@@ -545,8 +595,8 @@ class TestSearchRagForControl:
assert "diagnostic principal" in first_call_query
@patch("src.medical.rag_search.search_similar_cpam")
def test_max_10_results(self, mock_search):
"""Le résultat final est limité à 10 entrées."""
def test_max_12_results(self, mock_search):
"""Le résultat final est limité à 12 entrées."""
mock_search.return_value = [
{"document": "guide_methodo", "page": i, "code": None,
"score": 0.9 - i * 0.01, "extrait": f"Texte {i}"}
@@ -558,7 +608,67 @@ class TestSearchRagForControl:
results = _search_rag_for_control(controle, dossier)
assert len(results) <= 10
assert len(results) <= 12
@patch("src.medical.rag_search.search_similar_cpam")
def test_arg_ucr_not_truncated_200(self, mock_search):
"""La requête RAG argument utilise jusqu'à 500 chars, pas 200."""
mock_search.return_value = []
dossier = _make_dossier()
long_arg = "A" * 400
controle = ControleCPAM(
numero_ogc=1, titre="Test", arg_ucr=long_arg,
decision_ucr="Rejet", dp_ucr=None, da_ucr=None,
)
_search_rag_for_control(controle, dossier)
# La requête argument doit contenir les 400 chars (pas tronquée à 200)
arg_call_query = mock_search.call_args_list[0][0][0]
assert len(arg_call_query) > 200
@patch("src.control.cpam_response.validate_code", return_value=(True, "Iléus paralytique"))
@patch("src.control.cpam_response.normalize_code", return_value="K56.0")
@patch("src.medical.rag_search.search_similar_cpam")
def test_query_cim10_definitions(self, mock_search, mock_norm, mock_valid):
"""Requête 4 exécutée quand codes contestés présents."""
mock_search.return_value = []
dossier = _make_dossier()
controle = _make_controle() # da_ucr="K56.0"
_search_rag_for_control(controle, dossier)
# Chercher la requête contenant "CIM-10" et "définition"
cim10_queries = [
c[0][0] for c in mock_search.call_args_list
if "CIM-10" in c[0][0] and "définition" in c[0][0]
]
assert len(cim10_queries) >= 1
assert "K56.0" in cim10_queries[0]
@patch("src.medical.rag_search.search_similar_cpam")
def test_query_rule_extraction(self, mock_search):
"""Requête 5 exécutée quand arg_ucr contient une règle nommée."""
mock_search.return_value = []
dossier = _make_dossier()
controle = ControleCPAM(
numero_ogc=1, titre="Désaccord DAS",
arg_ucr="Selon la RègleT7 et l'Annexe-4B, le DAS n'est pas justifié.",
decision_ucr="Rejet", dp_ucr=None, da_ucr=None,
)
_search_rag_for_control(controle, dossier)
# Chercher la requête contenant les règles extraites
rule_queries = [
c[0][0] for c in mock_search.call_args_list
if "guide méthodologique" in c[0][0]
]
assert len(rule_queries) >= 1
assert "RègleT7" in rule_queries[0] or "Annexe" in rule_queries[0]
@patch("src.medical.rag_search.search_similar_cpam")
def test_clinical_query_when_das_match(self, mock_search):
@@ -570,8 +680,11 @@ class TestSearchRagForControl:
_search_rag_for_control(controle, dossier)
# 3 appels : codes + argument + clinique
assert mock_search.call_count == 3
third_call_query = mock_search.call_args_list[2][0][0]
assert "Iléus réflexe" in third_call_query
assert "Cholécystite aiguë" in third_call_query
# Au moins 4 appels : codes + argument + clinique + CIM-10 définitions
assert mock_search.call_count >= 4
# La requête clinique contient DP + DAS textes
clinique_queries = [
c[0][0] for c in mock_search.call_args_list
if "Iléus réflexe" in c[0][0] and "Cholécystite aiguë" in c[0][0]
]
assert len(clinique_queries) >= 1

View File

@@ -2,7 +2,7 @@
import pytest
from src.viewer.app import create_app, compute_group_stats, severity_badge, format_duration
from src.viewer.app import create_app, compute_group_stats, severity_badge, format_duration, format_cpam_text
from src.config import DossierMedical, Diagnostic, ActeCCAM
@@ -104,6 +104,40 @@ class TestIndexPageLoads:
assert b"Dossiers" in response.data
class TestFormatCpamText:
def test_plain_text(self):
result = format_cpam_text("Un simple paragraphe.")
assert "<p" in result
assert "Un simple paragraphe." in result
def test_bullet_list(self):
result = format_cpam_text("- Premier argument\n- Deuxième argument")
assert "<ul" in result
assert "<li>Premier argument</li>" in result
assert "<li>Deuxième argument</li>" in result
def test_mixed_text_and_bullets(self):
text = "Introduction\n- Point A\n- Point B\nConclusion"
result = format_cpam_text(text)
assert "<p" in result
assert "<ul" in result
assert "<li>Point A</li>" in result
assert "Conclusion" in result
def test_none_input(self):
result = format_cpam_text(None)
assert result == ""
def test_empty_input(self):
result = format_cpam_text("")
assert result == ""
def test_html_escaping(self):
result = format_cpam_text("Test <script>alert('xss')</script>")
assert "<script>" not in result
assert "&lt;script&gt;" in result
class TestDetailPageLoads:
def test_detail_page_404(self, client):
"""Un fichier inexistant retourne 404."""