feat: pipeline T2A - anonymisation, extraction CIM-10 et intégration edsnlp

Pipeline complet de traitement de documents médicaux PDF :
- Extraction texte (pdfplumber) et classification (Trackare/CRH)
- Anonymisation multi-couche (regex + NER CamemBERT + sweep)
- Extraction médicale CIM-10 hybride : edsnlp (AP-HP) enrichit les
  diagnostics, médicaments (codes ATC via Romedi) et négation,
  avec fallback regex pour les patterns spécifiques
- Fix sentencepiece pinné à <0.2.0 pour compatibilité CamemBERT

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

tests/test_anonymization.py

@@ -0,0 +1,197 @@
+"""Tests pour le module d'anonymisation."""
+
+import pytest
+
+from src.anonymization.entity_registry import EntityRegistry
+from src.anonymization.regex_patterns import (
+    CRH_FOOTER_IPP_EPISODE,
+    CRH_FOOTER_PATIENT_PATTERN,
+    DATE_NAISSANCE_PATTERN,
+    DR_NAME_PATTERN,
+    EMAIL_PATTERN,
+    EPISODE_PATTERN,
+    FOOTER_PATIENT_PATTERN,
+    IPP_PATTERN,
+    NOTE_AUTHOR_PATTERN,
+    PHONE_PATTERN,
+    RPPS_PATTERN,
+)
+
+
+class TestRegexPatterns:
+    def test_ipp_with_colon(self):
+        m = IPP_PATTERN.search("IPP: 01306172")
+        assert m is not None
+        assert m.group(1) == "01306172"
+
+    def test_ipp_without_colon(self):
+        m = IPP_PATTERN.search("IPP 01306172")
+        assert m is not None
+        assert m.group(1) == "01306172"
+
+    def test_ipp_in_parentheses(self):
+        m = IPP_PATTERN.search("(01306172 )")
+        assert m is not None
+        assert m.group(2) == "01306172"
+
+    def test_episode_no(self):
+        m = EPISODE_PATTERN.search("Episode No: 23042753")
+        assert m is not None
+        assert m.group(1) == "23042753"
+
+    def test_episode_n_degree(self):
+        m = EPISODE_PATTERN.search("N° Episode 23042753")
+        assert m is not None
+        assert m.group(2) == "23042753"
+
+    def test_phone_dots(self):
+        m = PHONE_PATTERN.search("06.25.39.26.82")
+        assert m is not None
+        assert m.group(0) == "06.25.39.26.82"
+
+    def test_phone_spaces(self):
+        m = PHONE_PATTERN.search("05 59 44 35 35")
+        assert m is not None
+
+    def test_email(self):
+        m = EMAIL_PATTERN.search("faudemar@ch-cotebasque.fr")
+        assert m is not None
+        assert m.group(0) == "faudemar@ch-cotebasque.fr"
+
+    def test_rpps(self):
+        m = RPPS_PATTERN.search("RPPS : 10100532760")
+        assert m is not None
+        assert m.group(1) == "10100532760"
+
+    def test_date_naissance_nee_le(self):
+        m = DATE_NAISSANCE_PATTERN.search("née le 23/02/1980")
+        assert m is not None
+        assert m.group(1) == "23/02/1980"
+
+    def test_date_naissance_ne_e_le(self):
+        m = DATE_NAISSANCE_PATTERN.search("Né(e) le 23/02/1980")
+        assert m is not None
+        assert m.group(1) == "23/02/1980"
+
+    def test_date_naissance_field(self):
+        m = DATE_NAISSANCE_PATTERN.search("Date de naissance: 23/02/1980")
+        assert m is not None
+        assert m.group(1) == "23/02/1980"
+
+    def test_dr_name(self):
+        m = DR_NAME_PATTERN.search("Dr F. AUDEMAR")
+        assert m is not None
+        assert "AUDEMAR" in m.group(1)
+
+    def test_dr_name_docteur(self):
+        m = DR_NAME_PATTERN.search("Docteur AUDEMAR Franck")
+        assert m is not None
+        assert "AUDEMAR" in m.group(1)
+
+    def test_note_author(self):
+        m = NOTE_AUTHOR_PATTERN.search("Note IDE Annie GUIRESSE Non algique")
+        assert m is not None
+        assert m.group(1) == "Annie GUIRESSE"
+
+    def test_footer_patient_trackare(self):
+        m = FOOTER_PATIENT_PATTERN.search(
+            "Patient: CLIER NARBAIS AUDREY - Date de naissance: 23/02/1980"
+        )
+        assert m is not None
+        assert "CLIER" in m.group(1)
+
+    def test_crh_footer_patient(self):
+        m = CRH_FOOTER_PATIENT_PATTERN.search(
+            "Patient(e) : CLIER AUDREY NARBAIS Né(e) le 23/02/1980"
+        )
+        assert m is not None
+        assert "CLIER" in m.group(1)
+
+    def test_crh_footer_ipp_episode(self):
+        m = CRH_FOOTER_IPP_EPISODE.search(
+            "IPP 01306172 / N° Episode 23042753 (MEDECINE GASTRO B2 HC)"
+        )
+        assert m is not None
+        assert m.group(1) == "01306172"
+        assert m.group(2) == "23042753"
+
+
+class TestEntityRegistry:
+    def test_register_returns_pseudo(self):
+        reg = EntityRegistry()
+        pseudo = reg.register("Jean Dupont", "patient")
+        assert pseudo == "[PATIENT_1]"
+
+    def test_register_same_entity_returns_same(self):
+        reg = EntityRegistry()
+        p1 = reg.register("Jean Dupont", "patient")
+        p2 = reg.register("Jean Dupont", "patient")
+        assert p1 == p2
+
+    def test_register_case_insensitive(self):
+        reg = EntityRegistry()
+        p1 = reg.register("Jean DUPONT", "patient")
+        p2 = reg.register("jean dupont", "patient")
+        assert p1 == p2
+
+    def test_register_different_categories(self):
+        reg = EntityRegistry()
+        p1 = reg.register("Dupont", "patient")
+        p2 = reg.register("Martin", "medecin")
+        assert p1 == "[PATIENT_1]"
+        assert p2 == "[MEDECIN_1]"
+
+    def test_get_replacement(self):
+        reg = EntityRegistry()
+        reg.register("Jean Dupont", "patient")
+        assert reg.get_replacement("jean dupont") == "[PATIENT_1]"
+        assert reg.get_replacement("inconnu") is None
+
+
+class TestAnonymizer:
+    def test_anonymize_basic(self):
+        from src.anonymization.anonymizer import Anonymizer
+
+        parsed = {
+            "patient": {"nom_prenom": "DUPONT Jean", "nom_naissance": "DUPONT"},
+            "medecins": ["MARTIN Pierre"],
+            "contacts": [],
+        }
+        anonymizer = Anonymizer(parsed_data=parsed)
+        text = "Le patient DUPONT Jean a été vu par Dr MARTIN Pierre."
+        result = anonymizer.anonymize(text)
+
+        assert "DUPONT" not in result
+        assert "MARTIN" not in result
+        assert "[PATIENT" in result or "[MEDECIN" in result
+
+    def test_preserves_medical_content(self):
+        from src.anonymization.anonymizer import Anonymizer
+
+        anonymizer = Anonymizer(parsed_data={"patient": {}, "medecins": [], "contacts": []})
+        text = "Pancréatite aiguë biliaire. Cholécystectomie par cœlioscopie. IMC 34.37."
+        result = anonymizer.anonymize(text)
+
+        assert "Pancréatite" in result
+        assert "Cholécystectomie" in result
+        assert "IMC" in result
+
+    def test_anonymize_phone(self):
+        from src.anonymization.anonymizer import Anonymizer
+
+        anonymizer = Anonymizer(parsed_data={"patient": {}, "medecins": [], "contacts": []})
+        text = "Appeler le 06.25.39.26.82 pour le rendez-vous."
+        result = anonymizer.anonymize(text)
+
+        assert "06.25.39.26.82" not in result
+        assert "[TEL" in result
+
+    def test_anonymize_email(self):
+        from src.anonymization.anonymizer import Anonymizer
+
+        anonymizer = Anonymizer(parsed_data={"patient": {}, "medecins": [], "contacts": []})
+        text = "Contact: faudemar@ch-cotebasque.fr"
+        result = anonymizer.anonymize(text)
+
+        assert "faudemar@ch-cotebasque.fr" not in result
+        assert "[EMAIL" in result