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>
2026-02-17 23:24:10 +01:00
parent 94fa4e5f3b
commit bc0ccbef7c
7 changed files with 464 additions and 51 deletions
--- a/tests/test_cpam_response.py
+++ b/tests/test_cpam_response.py
@@ -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
--- a/tests/test_viewer.py
+++ b/tests/test_viewer.py
@@ -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."""