Externalize dictionaries and add anonymization review corpus

tests/synthetic_regression/manifest.json

@@ -0,0 +1,110 @@
+{
+  "001_patient_header_and_birth": {
+    "description": "En-tête patient en majuscules avec date de naissance masquée et date de soin conservée.",
+    "must_contain": [
+      "[DATE_NAISSANCE]",
+      "Consultation du 14/03/2024"
+    ],
+    "must_not_contain": [
+      "ETCHEVERRY",
+      "JEAN",
+      "CLAUDE",
+      "12/03/1980"
+    ]
+  },
+  "002_contact_bundle": {
+    "description": "Email et téléphone dans une même ligne de contact.",
+    "must_contain": [
+      "[EMAIL]",
+      "[TEL]"
+    ],
+    "must_not_contain": [
+      "jean.dupont@example.com",
+      "01 23 45 67 89"
+    ]
+  },
+  "003_multiline_venue_number": {
+    "description": "Numéro de venue éclaté sur deux lignes.",
+    "must_contain": [
+      "N° venue :",
+      "[NDA]",
+      "Date de séjour : 14/03/2024"
+    ],
+    "must_not_contain": [
+      "1234567"
+    ]
+  },
+  "004_identifier_bundle": {
+    "description": "Bloc d'identifiants structurés variés.",
+    "must_contain": [
+      "[RPPS]",
+      "[FINESS]",
+      "[IPP]",
+      "[OGC]",
+      "[IBAN]"
+    ],
+    "must_not_contain": [
+      "12345678901",
+      "123456789",
+      "ABC12345",
+      "FR76 3000 6000 0112 3456 7890 189"
+    ]
+  },
+  "005_force_mask_default_term": {
+    "description": "Terme forcé par la configuration par défaut.",
+    "must_contain": [
+      "[MASK]"
+    ],
+    "must_not_contain": [
+      "CHCB"
+    ]
+  },
+  "006_whitelist_phrases_preserved": {
+    "description": "Expressions métier explicitement préservées.",
+    "must_contain": [
+      "classification internationale",
+      "prise en charge"
+    ],
+    "must_not_contain": []
+  },
+  "007_overlay_force_mask_local": {
+    "description": "Terme local masqué via surcharge runtime.",
+    "must_contain": [
+      "[MASK]"
+    ],
+    "must_not_contain": [
+      "LOCAL_SIGLE"
+    ]
+  },
+  "008_ville_header": {
+    "description": "Ville en en-tête de courrier, date conservée.",
+    "must_contain": [
+      "[VILLE], le 12/03/2024"
+    ],
+    "must_not_contain": [
+      "Bayonne"
+    ]
+  },
+  "009_header_and_repeated_name": {
+    "description": "Propagation globale d'un nom vu dans l'en-tête.",
+    "must_contain": [
+      "Le patient [NOM] revient ce jour."
+    ],
+    "must_not_contain": [
+      "ETCHEVERRY",
+      "JEAN",
+      "CLAUDE"
+    ]
+  },
+  "010_spaced_establishment_header": {
+    "description": "En-tête d'établissement avec lettres espacées.",
+    "must_contain": [
+      "[ETABLISSEMENT]",
+      "Service de cardiologie"
+    ],
+    "must_not_contain": [
+      "C E N T R E",
+      "H O S P I T A L I E R"
+    ]
+  }
+}