diff --git a/src/config.py b/src/config.py
index 4cca12c..ee865ff 100644
--- a/src/config.py
+++ b/src/config.py
@@ -39,6 +39,7 @@ OLLAMA_TIMEOUT = 120
 
 RAG_INDEX_DIR = BASE_DIR / "data" / "rag_index"
 CIM10_DICT_PATH = BASE_DIR / "data" / "cim10_dict.json"
+CCAM_DICT_PATH = BASE_DIR / "data" / "ccam_dict.json"
 CIM10_PDF = Path("/home/dom/ai/aivanov_CIM/cim-10-fr_2026_a_usage_pmsi_version_provisoire_111225.pdf")
 GUIDE_METHODO_PDF = Path("/home/dom/ai/aivanov_CIM/guide_methodo_mco_2026_version_provisoire.pdf")
 CCAM_PDF = Path("/home/dom/ai/aivanov_CIM/actualisation_ccam_descriptive_a_usage_pmsi_v4_2025.pdf")
diff --git a/src/main.py b/src/main.py
index 21eb82e..41ae3b9 100644
--- a/src/main.py
+++ b/src/main.py
@@ -168,6 +168,18 @@ def main(input_path: str | None = None) -> None:
         action="store_true",
         help="Générer le dictionnaire CIM-10 depuis metadata.json et quitter",
     )
+    parser.add_argument(
+        "--build-ccam-dict",
+        nargs="?",
+        const="CCAM_V81.xls",
+        metavar="PATH",
+        help="Générer le dictionnaire CCAM depuis un fichier XLS (défaut: CCAM_V81.xls)",
+    )
+    parser.add_argument(
+        "--rebuild-index",
+        action="store_true",
+        help="Forcer la reconstruction de l'index FAISS",
+    )
     args = parser.parse_args()
 
     if args.build_dict:
@@ -175,6 +187,17 @@ def main(input_path: str | None = None) -> None:
         build_dict()
         return
 
+    if args.build_ccam_dict:
+        from .medical.ccam_dict import build_dict as build_ccam
+        result = build_ccam(args.build_ccam_dict)
+        logger.info("Dictionnaire CCAM : %d codes générés", len(result))
+        return
+
+    if args.rebuild_index:
+        from .medical.rag_index import build_index
+        build_index(force=True)
+        return
+
     if args.no_ner:
         # Monkey-patch pour désactiver NER
         from .anonymization import ner_anonymizer
diff --git a/src/medical/ccam_dict.py b/src/medical/ccam_dict.py
new file mode 100644
index 0000000..0200a20
--- /dev/null
+++ b/src/medical/ccam_dict.py
@@ -0,0 +1,191 @@
+"""Dictionnaire CCAM complet extrait depuis le fichier XLS officiel (CNAM).
+
+Fournit un lookup intelligent avec normalisation Unicode pour la recherche
+de codes CCAM à partir de textes d'actes médicaux en français.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import re
+import unicodedata
+from pathlib import Path
+from typing import Optional
+
+from ..config import CCAM_DICT_PATH
+
+logger = logging.getLogger(__name__)
+
+# Singleton : dictionnaire chargé une seule fois
+_dict_cache: dict[str, dict] | None = None
+# Cache des labels normalisés pour le substring matching
+_normalized_cache: list[tuple[str, str, str]] | None = None
+
+_CCAM_CODE_RE = re.compile(r"^[A-Z]{4}\d{3}$")
+
+
+def normalize_text(text: str) -> str:
+    """Normalise un texte : accent folding, lowercase, collapse whitespace."""
+    text = text.replace("\u2019", "'").replace("\u2018", "'").replace("\u02BC", "'")
+    nfkd = unicodedata.normalize("NFKD", text)
+    stripped = "".join(c for c in nfkd if unicodedata.category(c) != "Mn")
+    return re.sub(r"\s+", " ", stripped.lower()).strip()
+
+
+def build_dict(source_path: str | Path) -> dict[str, dict]:
+    """Construit le dictionnaire CCAM depuis un fichier XLS et l'écrit en JSON.
+
+    Format JSON : {code: {description, activite, tarif_s1, regroupement}}
+
+    Args:
+        source_path: Chemin vers le fichier XLS CCAM (ex: CCAM_V81.xls).
+
+    Returns:
+        Le dictionnaire code → infos.
+    """
+    import xlrd
+
+    source_path = Path(source_path)
+    if not source_path.exists():
+        logger.error("Fichier XLS non trouvé : %s", source_path)
+        return {}
+
+    wb = xlrd.open_workbook(str(source_path))
+    sheet = wb.sheet_by_index(0)
+
+    result: dict[str, dict] = {}
+
+    for r in range(sheet.nrows):
+        code = str(sheet.cell_value(r, 0)).strip()
+        if not _CCAM_CODE_RE.match(code):
+            continue
+
+        description = str(sheet.cell_value(r, 2)).strip()
+        activite_raw = sheet.cell_value(r, 3)
+        activite = int(activite_raw) if isinstance(activite_raw, float) else None
+
+        tarif_raw = sheet.cell_value(r, 5)
+        tarif_s1 = round(tarif_raw, 2) if isinstance(tarif_raw, (int, float)) else None
+
+        regroupement = str(sheet.cell_value(r, 10)).strip() or None
+
+        result[code] = {
+            "description": description,
+            "activite": activite,
+            "tarif_s1": tarif_s1,
+            "regroupement": regroupement,
+        }
+
+    # Écrire le fichier JSON
+    CCAM_DICT_PATH.parent.mkdir(parents=True, exist_ok=True)
+    with open(CCAM_DICT_PATH, "w", encoding="utf-8") as f:
+        json.dump(result, f, ensure_ascii=False, indent=2)
+
+    logger.info("Dictionnaire CCAM généré : %d codes → %s", len(result), CCAM_DICT_PATH)
+    return result
+
+
+def load_dict() -> dict[str, dict]:
+    """Charge le dictionnaire CCAM (singleton lazy-loaded).
+
+    Si le fichier JSON n'existe pas, retourne un dict vide avec un warning.
+    """
+    global _dict_cache
+    if _dict_cache is not None:
+        return _dict_cache
+
+    if CCAM_DICT_PATH.exists():
+        with open(CCAM_DICT_PATH, encoding="utf-8") as f:
+            _dict_cache = json.load(f)
+    else:
+        logger.warning("Dictionnaire CCAM absent : %s — lancez --build-ccam-dict", CCAM_DICT_PATH)
+        _dict_cache = {}
+
+    return _dict_cache
+
+
+def _get_normalized_entries() -> list[tuple[str, str, str]]:
+    """Retourne une liste de (code, description, description_normalisée) triée par longueur."""
+    global _normalized_cache
+    if _normalized_cache is not None:
+        return _normalized_cache
+
+    d = load_dict()
+    entries = []
+    for code, info in d.items():
+        desc = info.get("description", "") if isinstance(info, dict) else str(info)
+        norm = normalize_text(desc)
+        entries.append((code, desc, norm))
+
+    # Trier par longueur de description décroissante (plus spécifique d'abord)
+    entries.sort(key=lambda e: -len(e[2]))
+    _normalized_cache = entries
+    return _normalized_cache
+
+
+def lookup(
+    text: str,
+    domain_overrides: dict[str, str] | None = None,
+) -> str | None:
+    """Recherche un code CCAM pour un texte donné.
+
+    Stratégie en 3 niveaux :
+    1. Match substring dans domain_overrides (prioritaire, ex: CCAM_MAP existant)
+    2. Match exact normalisé dans le dictionnaire complet
+    3. Match substring normalisé avec scoring par spécificité
+
+    Args:
+        text: Le texte de l'acte médical à rechercher.
+        domain_overrides: Dictionnaire terme→code prioritaire.
+
+    Returns:
+        Le code CCAM trouvé ou None.
+    """
+    if not text:
+        return None
+
+    text_norm = normalize_text(text)
+
+    # Niveau 1 : domain overrides (substring match)
+    if domain_overrides:
+        for terme, code in domain_overrides.items():
+            if normalize_text(terme) in text_norm:
+                return code
+
+    entries = _get_normalized_entries()
+
+    # Niveau 2 : match exact normalisé
+    for code, _desc, norm_desc in entries:
+        if norm_desc == text_norm:
+            return code
+
+    # Niveau 3 : substring match normalisé (plus spécifique d'abord)
+    for code, _desc, norm_desc in entries:
+        if not norm_desc or len(norm_desc) < 4:
+            continue
+        if norm_desc in text_norm or text_norm in norm_desc:
+            return code
+
+    return None
+
+
+def validate_code(code: str) -> tuple[bool, str]:
+    """Vérifie si un code CCAM existe dans le dictionnaire.
+
+    Returns:
+        (is_valid, description) — description vide si invalide.
+    """
+    d = load_dict()
+    if code in d:
+        info = d[code]
+        desc = info.get("description", "") if isinstance(info, dict) else str(info)
+        return True, desc
+    return False, ""
+
+
+def reset_cache() -> None:
+    """Réinitialise les caches (utile pour les tests)."""
+    global _dict_cache, _normalized_cache
+    _dict_cache = None
+    _normalized_cache = None
diff --git a/src/medical/cim10_extractor.py b/src/medical/cim10_extractor.py
index f41d012..1932cdb 100644
--- a/src/medical/cim10_extractor.py
+++ b/src/medical/cim10_extractor.py
@@ -10,6 +10,7 @@ from typing import Optional
 logger = logging.getLogger(__name__)
 
 from .cim10_dict import lookup as dict_lookup, normalize_text
+from .ccam_dict import lookup as ccam_lookup, validate_code as ccam_validate
 from ..config import (
     ActeCCAM,
     BiologieCle,
@@ -113,6 +114,9 @@ def extract_medical_info(
     if use_rag:
         _enrich_with_rag(dossier)
 
+    # Post-processing : validation des codes CCAM contre le dictionnaire
+    _validate_ccam(dossier)
+
     # Post-processing : exclusions symptôme vs diagnostic précis
     _apply_exclusion_rules(dossier)
 
@@ -395,6 +399,13 @@ def _extract_actes(text: str, dossier: DossierMedical) -> None:
             date=date,
         ))
 
+    # Fallback : tenter le lookup CCAM dict pour les actes sans code
+    for acte in dossier.actes_ccam:
+        if not acte.code_ccam_suggestion:
+            code = ccam_lookup(acte.texte, domain_overrides=CCAM_MAP)
+            if code:
+                acte.code_ccam_suggestion = code
+
 
 def _extract_antecedents(text: str, dossier: DossierMedical) -> None:
     """Extrait les antécédents."""
@@ -625,6 +636,22 @@ def _is_negated_by_edsnlp(term: str, negated_terms: set[str]) -> bool:
     return False
 
 
+def _validate_ccam(dossier: DossierMedical) -> None:
+    """Valide les codes CCAM suggérés contre le dictionnaire officiel."""
+    for acte in dossier.actes_ccam:
+        if not acte.code_ccam_suggestion:
+            acte.validite = "non_verifie"
+            continue
+        is_valid, desc = ccam_validate(acte.code_ccam_suggestion)
+        if is_valid:
+            acte.validite = "valide"
+        else:
+            acte.validite = "non_verifie"
+            dossier.alertes_codage.append(
+                f"CCAM {acte.code_ccam_suggestion} ({acte.texte}) : code absent du dictionnaire CCAM V81"
+            )
+
+
 def _find_act_date(text: str, act_pattern: str) -> str | None:
     """Trouve la date associée à un acte."""
     # Chercher "acte le DD/MM" ou "acte le DD/MM/YYYY"
diff --git a/src/medical/rag_index.py b/src/medical/rag_index.py
index a59e323..6b7fbd8 100644
--- a/src/medical/rag_index.py
+++ b/src/medical/rag_index.py
@@ -11,7 +11,7 @@ from typing import Optional
 
 import pdfplumber
 
-from ..config import RAG_INDEX_DIR, CIM10_PDF, GUIDE_METHODO_PDF, CCAM_PDF
+from ..config import RAG_INDEX_DIR, CIM10_PDF, GUIDE_METHODO_PDF, CCAM_PDF, CCAM_DICT_PATH
 
 logger = logging.getLogger(__name__)
 
@@ -33,18 +33,46 @@ class Chunk:
 # ---------------------------------------------------------------------------
 
 def _chunk_cim10(pdf_path: Path) -> list[Chunk]:
-    """Découpe le PDF CIM-10 en chunks par code 3 caractères (ex: K80, K85)."""
+    """Découpe le PDF CIM-10 en double chunking : sous-codes individuels + parents 3-char."""
     chunks: list[Chunk] = []
-    current_code: str | None = None
-    current_text: list[str] = []
-    current_page: int | None = None
+    current_code3: str | None = None
+    current_code3_text: list[str] = []
+    current_code3_page: int | None = None
+
+    # Sous-codes en cours d'accumulation
+    current_subcode: str | None = None
+    current_subcode_text: list[str] = []
+    current_subcode_page: int | None = None
 
-    # Pattern pour détecter un code CIM-10 à 3 caractères en début de ligne
     code3_pattern = re.compile(r"^([A-Z]\d{2})\s+(.+)")
-    # Pattern pour les sous-codes (ex: K80.0, K80.1)
     subcode_pattern = re.compile(r"^([A-Z]\d{2}\.\d+)\s+(.+)")
 
-    logger.info("Extraction des chunks CIM-10 depuis %s", pdf_path.name)
+    logger.info("Extraction des chunks CIM-10 (double chunking) depuis %s", pdf_path.name)
+
+    def _flush_subcode():
+        """Sauvegarde le chunk sous-code en cours."""
+        if current_subcode and current_subcode_text:
+            chunk_text = "\n".join(current_subcode_text)
+            if len(chunk_text.split()) >= 3:
+                chunks.append(Chunk(
+                    text=chunk_text,
+                    document="cim10",
+                    page=current_subcode_page,
+                    code=current_subcode,
+                ))
+
+    def _flush_code3():
+        """Sauvegarde le chunk parent 3-char en cours."""
+        _flush_subcode()
+        if current_code3 and current_code3_text:
+            chunk_text = "\n".join(current_code3_text)
+            if len(chunk_text.split()) >= 5:
+                chunks.append(Chunk(
+                    text=chunk_text,
+                    document="cim10",
+                    page=current_code3_page,
+                    code=current_code3,
+                ))
 
     with pdfplumber.open(pdf_path) as pdf:
         for page_num, page in enumerate(pdf.pages, start=1):
@@ -57,37 +85,38 @@ def _chunk_cim10(pdf_path: Path) -> list[Chunk]:
                 if not line:
                     continue
 
-                m = code3_pattern.match(line)
-                if m and not subcode_pattern.match(line):
-                    # Nouveau code 3-char → sauvegarder le chunk précédent
-                    if current_code and current_text:
-                        chunk_text = "\n".join(current_text)
-                        if len(chunk_text.split()) >= 5:
-                            chunks.append(Chunk(
-                                text=chunk_text,
-                                document="cim10",
-                                page=current_page,
-                                code=current_code,
-                            ))
-                    current_code = m.group(1)
-                    current_text = [line]
-                    current_page = page_num
+                m_sub = subcode_pattern.match(line)
+                m3 = code3_pattern.match(line)
+
+                if m_sub:
+                    # Nouveau sous-code → flush le sous-code précédent
+                    _flush_subcode()
+                    current_subcode = m_sub.group(1)
+                    current_subcode_text = [line]
+                    current_subcode_page = page_num
+                    # Ajouter aussi au chunk parent
+                    if current_code3:
+                        current_code3_text.append(line)
+                elif m3 and not m_sub:
+                    # Nouveau code 3-char → flush tout le bloc précédent
+                    _flush_code3()
+                    current_code3 = m3.group(1)
+                    current_code3_text = [line]
+                    current_code3_page = page_num
+                    current_subcode = None
+                    current_subcode_text = []
+                    current_subcode_page = None
                 else:
-                    if current_code:
-                        current_text.append(line)
+                    # Ligne de continuation
+                    if current_subcode:
+                        current_subcode_text.append(line)
+                    if current_code3:
+                        current_code3_text.append(line)
 
-    # Dernier chunk
-    if current_code and current_text:
-        chunk_text = "\n".join(current_text)
-        if len(chunk_text.split()) >= 5:
-            chunks.append(Chunk(
-                text=chunk_text,
-                document="cim10",
-                page=current_page,
-                code=current_code,
-            ))
+    # Flush final
+    _flush_code3()
 
-    logger.info("CIM-10 : %d chunks extraits", len(chunks))
+    logger.info("CIM-10 : %d chunks extraits (double chunking sous-codes + parents)", len(chunks))
     return chunks
 
 
@@ -253,6 +282,95 @@ def _chunk_ccam(pdf_path: Path) -> list[Chunk]:
     return chunks
 
 
+# ---------------------------------------------------------------------------
+# Chunking CCAM depuis le dictionnaire JSON
+# ---------------------------------------------------------------------------
+
+def _chunk_ccam_from_dict() -> list[Chunk]:
+    """Génère des chunks CCAM depuis ccam_dict.json (un chunk par code+description).
+
+    Prioritaire sur les chunks PDF si le dictionnaire existe.
+    """
+    if not CCAM_DICT_PATH.exists():
+        return []
+
+    import json as _json
+    with open(CCAM_DICT_PATH, encoding="utf-8") as f:
+        ccam_dict = _json.load(f)
+
+    chunks: list[Chunk] = []
+    for code, info in ccam_dict.items():
+        desc = info.get("description", "") if isinstance(info, dict) else str(info)
+        if not desc:
+            continue
+        regroupement = info.get("regroupement", "") if isinstance(info, dict) else ""
+        tarif = info.get("tarif_s1") if isinstance(info, dict) else None
+        text_parts = [f"{code} {desc}"]
+        if regroupement:
+            text_parts.append(f"Regroupement: {regroupement}")
+        if tarif is not None:
+            text_parts.append(f"Tarif S1: {tarif}€")
+        chunks.append(Chunk(
+            text="\n".join(text_parts),
+            document="ccam",
+            code=code,
+        ))
+
+    logger.info("CCAM dict : %d chunks générés depuis %s", len(chunks), CCAM_DICT_PATH)
+    return chunks
+
+
+# ---------------------------------------------------------------------------
+# Chunking CIM-10 Index Alphabétique
+# ---------------------------------------------------------------------------
+
+def _chunk_cim10_alpha(pdf_path: Path) -> list[Chunk]:
+    """Parse la section INDEX ALPHABÉTIQUE du PDF CIM-10.
+
+    Détecte les entrées de type "terme → code" et génère des chunks
+    avec document="cim10_alpha".
+    """
+    chunks: list[Chunk] = []
+    # Pattern : ligne avec un terme suivi d'un code CIM-10 en fin de ligne
+    entry_pattern = re.compile(r"^(.+?)\s+([A-Z]\d{2}(?:\.\d+)?)\s*$")
+
+    logger.info("Extraction de l'index alphabétique CIM-10 depuis %s", pdf_path.name)
+
+    in_alpha_section = False
+    with pdfplumber.open(pdf_path) as pdf:
+        for page_num, page in enumerate(pdf.pages, start=1):
+            text = page.extract_text()
+            if not text:
+                continue
+
+            # Détecter le début de la section index alphabétique
+            text_upper = text.upper()
+            if "INDEX ALPHAB" in text_upper:
+                in_alpha_section = True
+            # Certaines pages avant l'index : ne pas parser
+            if not in_alpha_section:
+                continue
+
+            for line in text.split("\n"):
+                line = line.strip()
+                if not line:
+                    continue
+                m = entry_pattern.match(line)
+                if m:
+                    terme = m.group(1).strip()
+                    code = m.group(2)
+                    if len(terme) >= 3:
+                        chunks.append(Chunk(
+                            text=f"{terme} → {code}",
+                            document="cim10_alpha",
+                            page=page_num,
+                            code=code,
+                        ))
+
+    logger.info("CIM-10 index alphabétique : %d entrées extraites", len(chunks))
+    return chunks
+
+
 # ---------------------------------------------------------------------------
 # Construction de l'index FAISS
 # ---------------------------------------------------------------------------
@@ -280,13 +398,25 @@ def build_index(force: bool = False) -> None:
     for pdf_path, chunk_fn in [
         (CIM10_PDF, _chunk_cim10),
         (GUIDE_METHODO_PDF, _chunk_guide_methodo),
-        (CCAM_PDF, _chunk_ccam),
     ]:
         if pdf_path.exists():
             all_chunks.extend(chunk_fn(pdf_path))
         else:
             logger.warning("PDF non trouvé : %s", pdf_path)
 
+    # CCAM : priorité au dictionnaire JSON sur le PDF
+    ccam_dict_chunks = _chunk_ccam_from_dict()
+    if ccam_dict_chunks:
+        all_chunks.extend(ccam_dict_chunks)
+    elif CCAM_PDF.exists():
+        all_chunks.extend(_chunk_ccam(CCAM_PDF))
+    else:
+        logger.warning("Ni dictionnaire CCAM ni PDF CCAM trouvé")
+
+    # CIM-10 index alphabétique (source additionnelle)
+    if CIM10_PDF.exists():
+        all_chunks.extend(_chunk_cim10_alpha(CIM10_PDF))
+
     if not all_chunks:
         logger.error("Aucun chunk extrait — vérifiez les chemins des PDFs")
         return
@@ -316,9 +446,9 @@ def build_index(force: bool = False) -> None:
 
     metadata = [asdict(c) for c in all_chunks]
     # Ne pas sauvegarder le texte complet dans metadata (trop lourd),
-    # garder un extrait de 500 chars
+    # garder un extrait de 800 chars (les sous-codes sont courts, besoin du contexte)
     for m in metadata:
-        m["extrait"] = m.pop("text")[:500]
+        m["extrait"] = m.pop("text")[:800]
 
     meta_path.write_text(json.dumps(metadata, ensure_ascii=False, indent=2), encoding="utf-8")
 
diff --git a/src/medical/rag_search.py b/src/medical/rag_search.py
index b028fba..1b42acc 100644
--- a/src/medical/rag_search.py
+++ b/src/medical/rag_search.py
@@ -74,8 +74,8 @@ def search_similar(query: str, top_k: int = 10) -> list[dict]:
         raw_results.append(meta)
 
     # Prioriser les sources CIM-10 (au moins 6 sur top_k)
-    cim10_results = [r for r in raw_results if r["document"] == "cim10"]
-    other_results = [r for r in raw_results if r["document"] != "cim10"]
+    cim10_results = [r for r in raw_results if r["document"] in ("cim10", "cim10_alpha")]
+    other_results = [r for r in raw_results if r["document"] not in ("cim10", "cim10_alpha")]
 
     min_cim10 = min(6, len(cim10_results))
     final = cim10_results[:min_cim10]
@@ -150,6 +150,7 @@ def _build_prompt(texte: str, sources: list[dict], contexte: dict, est_dp: bool
     for i, src in enumerate(sources, 1):
         doc_name = {
             "cim10": "CIM-10 FR 2026",
+            "cim10_alpha": "CIM-10 Index Alphabétique 2026",
             "guide_methodo": "Guide Méthodologique MCO 2026",
             "ccam": "CCAM PMSI V4 2025",
         }.get(src["document"], src["document"])
diff --git a/src/viewer/app.py b/src/viewer/app.py
index 5ec0c7d..e1f6aa2 100644
--- a/src/viewer/app.py
+++ b/src/viewer/app.py
@@ -147,4 +147,37 @@ def create_app() -> Flask:
         logger.info("Modèle Ollama changé : %s", new_model)
         return jsonify({"ok": True, "model": cfg.OLLAMA_MODEL})
 
+    @app.route("/reprocess/<path:filepath>", methods=["POST"])
+    def reprocess(filepath: str):
+        """Relance le traitement d'un dossier."""
+        from ..main import process_pdf, write_outputs
+        
+        dossier = load_dossier(filepath)
+        source_file = dossier.source_file
+        if not source_file:
+            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
+        
+        if not pdf_path:
+            return jsonify({"error": f"PDF source '{source_file}' introuvable"}), 404
+        
+        try:
+            anonymized_text, new_dossier, report = process_pdf(pdf_path)
+            stem = pdf_path.stem.replace(" ", "_")
+            subdir = None
+            if pdf_path.parent != input_dir:
+                subdir = pdf_path.parent.name
+            write_outputs(stem, anonymized_text, new_dossier, report, subdir=subdir)
+            return jsonify({"ok": True, "message": "Traitement terminé"})
+        except Exception as e:
+            logger.exception("Erreur lors du retraitement")
+            return jsonify({"error": str(e)}), 500
+
     return app
diff --git a/src/viewer/templates/base.html b/src/viewer/templates/base.html
index 7f1db19..b40dd0b 100644
--- a/src/viewer/templates/base.html
+++ b/src/viewer/templates/base.html
@@ -253,6 +253,7 @@
 
   loadModels();
 })();
+{% block scripts %}{% endblock %}
 </script>
 </body>
 </html>
diff --git a/tests/test_ccam_dict.py b/tests/test_ccam_dict.py
new file mode 100644
index 0000000..09a1d65
--- /dev/null
+++ b/tests/test_ccam_dict.py
@@ -0,0 +1,113 @@
+"""Tests pour le dictionnaire CCAM (build, load, lookup, validate)."""
+
+from __future__ import annotations
+
+from pathlib import Path
+from unittest.mock import patch
+
+import pytest
+
+from src.medical.ccam_dict import (
+    build_dict,
+    load_dict,
+    lookup,
+    normalize_text,
+    reset_cache,
+    validate_code,
+)
+
+# Chemin vers le XLS de test (dans le repo)
+CCAM_XLS = Path(__file__).resolve().parent.parent / "CCAM_V81.xls"
+
+
+@pytest.fixture(autouse=True)
+def _clear_cache():
+    """Réinitialise le cache avant chaque test."""
+    reset_cache()
+    yield
+    reset_cache()
+
+
+@pytest.mark.skipif(not CCAM_XLS.exists(), reason="CCAM_V81.xls non trouvé")
+class TestBuildDict:
+    def test_build_dict_from_xls(self, tmp_path):
+        """Parsing du XLS → nombre de codes >= 8000."""
+        out = tmp_path / "ccam_dict.json"
+        with patch("src.medical.ccam_dict.CCAM_DICT_PATH", out):
+            result = build_dict(CCAM_XLS)
+        assert len(result) >= 8000, f"Seulement {len(result)} codes extraits"
+
+    def test_known_codes_present(self, tmp_path):
+        """HMFC004 (cholécystectomie) et ZCQK002 (radio abdo) doivent être présents."""
+        out = tmp_path / "ccam_dict.json"
+        with patch("src.medical.ccam_dict.CCAM_DICT_PATH", out):
+            result = build_dict(CCAM_XLS)
+        assert "HMFC004" in result, "HMFC004 (cholécystectomie) absent"
+        assert "ZCQK002" in result, "ZCQK002 (radio abdomen) absent"
+        assert "cholécystectomie" in result["HMFC004"]["description"].lower()
+
+
+@pytest.mark.skipif(not CCAM_XLS.exists(), reason="CCAM_V81.xls non trouvé")
+class TestLoadDict:
+    def test_load_dict_singleton(self, tmp_path):
+        """Chargement lazy + cache (le 2e appel retourne le même objet)."""
+        out = tmp_path / "ccam_dict.json"
+        with patch("src.medical.ccam_dict.CCAM_DICT_PATH", out):
+            build_dict(CCAM_XLS)
+        with patch("src.medical.ccam_dict.CCAM_DICT_PATH", out):
+            d1 = load_dict()
+            d2 = load_dict()
+        assert d1 is d2, "Le cache singleton ne fonctionne pas"
+        assert len(d1) >= 8000
+
+
+@pytest.mark.skipif(not CCAM_XLS.exists(), reason="CCAM_V81.xls non trouvé")
+class TestLookup:
+    @pytest.fixture(autouse=True)
+    def _build(self, tmp_path):
+        out = tmp_path / "ccam_dict.json"
+        with patch("src.medical.ccam_dict.CCAM_DICT_PATH", out):
+            build_dict(CCAM_XLS)
+        # Charger dans le cache
+        with patch("src.medical.ccam_dict.CCAM_DICT_PATH", out):
+            load_dict()
+
+    def test_lookup_exact(self):
+        """Lookup 'cholécystectomie' → doit trouver un code contenant ce terme."""
+        code = lookup("Cholécystectomie, par cœlioscopie")
+        assert code == "HMFC004", f"Attendu HMFC004, obtenu {code}"
+
+    def test_lookup_substring(self):
+        """Lookup 'cholécystectomie par cœlioscopie' → HMFC004."""
+        code = lookup("cholécystectomie")
+        assert code is not None
+        # Doit matcher un code contenant "cholécystectomie"
+        assert code == "HMFC004" or code is not None
+
+    def test_lookup_unknown(self):
+        """Un texte totalement hors domaine retourne None."""
+        code = lookup("xyz totalement inconnu blabla")
+        assert code is None
+
+
+@pytest.mark.skipif(not CCAM_XLS.exists(), reason="CCAM_V81.xls non trouvé")
+class TestValidateCode:
+    @pytest.fixture(autouse=True)
+    def _build(self, tmp_path):
+        out = tmp_path / "ccam_dict.json"
+        with patch("src.medical.ccam_dict.CCAM_DICT_PATH", out):
+            build_dict(CCAM_XLS)
+        with patch("src.medical.ccam_dict.CCAM_DICT_PATH", out):
+            load_dict()
+
+    def test_validate_code_known(self):
+        """HMFC004 → valide."""
+        is_valid, desc = validate_code("HMFC004")
+        assert is_valid is True
+        assert "cholécystectomie" in desc.lower()
+
+    def test_validate_code_unknown(self):
+        """XXXXX99 → invalide."""
+        is_valid, desc = validate_code("XXXXX99")
+        assert is_valid is False
+        assert desc == ""
diff --git a/tests/test_rag.py b/tests/test_rag.py
index 1b9fc0c..9339783 100644
--- a/tests/test_rag.py
+++ b/tests/test_rag.py
@@ -44,6 +44,7 @@ class TestDiagnosticExtended:
         assert d.cim10_suggestion == "K85.9"
         assert d.cim10_confidence is None
         assert d.justification is None
+        assert d.raisonnement is None
         assert d.sources_rag == []
 
     def test_with_rag_fields(self):
@@ -52,12 +53,15 @@ class TestDiagnosticExtended:
             cim10_suggestion="K80.5",
             cim10_confidence="high",
             justification="Code K80.5 correspond à la lithiase du cholédoque",
+            raisonnement="1. ANALYSE CLINIQUE : La lithiase cholédoque est...",
             sources_rag=[
                 RAGSource(document="cim10", page=480, code="K80"),
             ],
         )
         assert d.cim10_confidence == "high"
         assert d.justification is not None
+        assert d.raisonnement is not None
+        assert d.raisonnement.startswith("1. ANALYSE CLINIQUE")
         assert len(d.sources_rag) == 1
         assert d.sources_rag[0].code == "K80"
 
@@ -67,6 +71,7 @@ class TestDiagnosticExtended:
         data = d.model_dump(exclude_none=True)
         assert "cim10_confidence" not in data
         assert "justification" not in data
+        assert "raisonnement" not in data
         assert "sources_rag" in data  # list vide incluse
 
     def test_dossier_with_extended_diagnostic(self):
@@ -77,6 +82,7 @@ class TestDiagnosticExtended:
                 cim10_suggestion="K85.1",
                 cim10_confidence="high",
                 justification="Confirmé par CIM-10 FR 2026",
+                raisonnement="Le DP K85.1 est le code le plus spécifique...",
                 sources_rag=[
                     RAGSource(document="cim10", page=496, code="K85"),
                     RAGSource(document="guide_methodo", page=30),
@@ -84,6 +90,7 @@ class TestDiagnosticExtended:
             ),
         )
         assert dossier.diagnostic_principal.cim10_confidence == "high"
+        assert dossier.diagnostic_principal.raisonnement is not None
         assert len(dossier.diagnostic_principal.sources_rag) == 2
 
 
@@ -152,10 +159,32 @@ class TestChunkingCIM10:
         assert len(chunks) > 100, f"Trop peu de chunks : {len(chunks)}"
 
         codes = {c.code for c in chunks if c.code}
+        # Codes parents 3-char
         assert "K85" in codes, "K85 (pancréatite) non trouvé"
         assert "K80" in codes, "K80 (lithiase biliaire) non trouvé"
         assert "E66" in codes, "E66 (obésité) non trouvé"
 
+    @pytest.mark.skipif(
+        not CIM10_PDF.exists(),
+        reason=f"PDF CIM-10 non trouvé : {CIM10_PDF}",
+    )
+    def test_double_chunking_subcodes(self):
+        """Le double chunking produit des chunks sous-codes (X99.9) en plus des parents."""
+        from src.medical.rag_index import _chunk_cim10
+
+        chunks = _chunk_cim10(CIM10_PDF)
+        codes = {c.code for c in chunks if c.code}
+
+        # Il doit y avoir des sous-codes (avec un point)
+        subcodes = {c for c in codes if "." in c}
+        assert len(subcodes) > 100, f"Trop peu de sous-codes : {len(subcodes)}"
+
+        # Le nombre total de chunks doit être significativement plus grand
+        # qu'un chunking simple par code 3-char
+        parent_codes = {c for c in codes if "." not in c}
+        assert len(chunks) > len(parent_codes) * 2, \
+            f"Double chunking inefficace : {len(chunks)} chunks pour {len(parent_codes)} codes parents"
+
     @pytest.mark.skipif(
         not CIM10_PDF.exists(),
         reason=f"PDF CIM-10 non trouvé : {CIM10_PDF}",
@@ -164,9 +193,10 @@ class TestChunkingCIM10:
         from src.medical.rag_index import _chunk_cim10
 
         chunks = _chunk_cim10(CIM10_PDF)
-        k85_chunks = [c for c in chunks if c.code == "K85"]
-        assert len(k85_chunks) >= 1
-        assert "pancréatite" in k85_chunks[0].text.lower() or "pancreatite" in k85_chunks[0].text.lower()
+        k85_chunks = [c for c in chunks if c.code and c.code.startswith("K85")]
+        assert len(k85_chunks) >= 2, "Il devrait y avoir au moins un chunk parent K85 + des sous-codes"
+        texts_lower = " ".join(c.text.lower() for c in k85_chunks)
+        assert "pancréatite" in texts_lower or "pancreatite" in texts_lower
 
 
 class TestChunkingGuideMethodo:
@@ -195,6 +225,183 @@ class TestChunkingCCAM:
         assert all(c.document == "ccam" for c in chunks)
 
 
+class TestParseOllamaResponse:
+    """Tests pour _parse_ollama_response avec le marqueur ###RESULT###."""
+
+    def test_parse_with_marker(self):
+        from src.medical.rag_search import _parse_ollama_response
+
+        raw = """1. ANALYSE CLINIQUE : La pancréatite aiguë biliaire est une inflammation...
+2. CODES CANDIDATS : K85.0, K85.1, K85.9
+3. DISCRIMINATION : K85.1 est spécifique à l'origine biliaire
+4. RÈGLE PMSI : Conforme pour un DP
+
+###RESULT###
+{"code": "K85.1", "confidence": "high", "justification": "Pancréatite aiguë d'origine biliaire"}"""
+
+        result = _parse_ollama_response(raw)
+        assert result is not None
+        assert result["code"] == "K85.1"
+        assert result["confidence"] == "high"
+        assert result["justification"] == "Pancréatite aiguë d'origine biliaire"
+        assert "raisonnement" in result
+        assert "ANALYSE CLINIQUE" in result["raisonnement"]
+
+    def test_parse_without_marker_fallback(self):
+        """Fallback sur la recherche d'accolades quand le marqueur est absent."""
+        from src.medical.rag_search import _parse_ollama_response
+
+        raw = """Voici mon analyse...
+{"code": "E66.0", "confidence": "medium", "justification": "Obésité due à un excès calorique"}"""
+
+        result = _parse_ollama_response(raw)
+        assert result is not None
+        assert result["code"] == "E66.0"
+        assert result["confidence"] == "medium"
+
+    def test_parse_empty_response(self):
+        from src.medical.rag_search import _parse_ollama_response
+
+        result = _parse_ollama_response("")
+        assert result is None
+
+    def test_parse_no_json(self):
+        from src.medical.rag_search import _parse_ollama_response
+
+        result = _parse_ollama_response("Réponse sans aucun JSON valide.")
+        assert result is None
+
+    def test_parse_invalid_json(self):
+        from src.medical.rag_search import _parse_ollama_response
+
+        raw = """###RESULT###
+{code: K85.1, invalid json}"""
+        result = _parse_ollama_response(raw)
+        assert result is None
+
+    def test_parse_marker_with_raisonnement_containing_braces(self):
+        """Le raisonnement peut contenir des accolades (ex: listes, exemples)."""
+        from src.medical.rag_search import _parse_ollama_response
+
+        raw = """Le code {K85} est un code parent.
+Sous-codes : {K85.0, K85.1, K85.2, K85.3}
+
+###RESULT###
+{"code": "K85.1", "confidence": "high", "justification": "Biliaire confirmé"}"""
+
+        result = _parse_ollama_response(raw)
+        assert result is not None
+        assert result["code"] == "K85.1"
+        assert "raisonnement" in result
+        assert "{K85}" in result["raisonnement"]
+
+
+class TestBuildPrompt:
+    """Tests pour le nouveau _build_prompt avec raisonnement structuré."""
+
+    def test_prompt_contains_diagnostic(self):
+        from src.medical.rag_search import _build_prompt
+
+        sources = [{"document": "cim10", "code": "K85", "page": 1, "extrait": "K85 Pancréatite"}]
+        contexte = {"sexe": "F", "age": 43}
+        prompt = _build_prompt("Pancréatite aiguë biliaire", sources, contexte, est_dp=True)
+
+        assert "Pancréatite aiguë biliaire" in prompt
+        assert "DP (diagnostic principal)" in prompt
+        assert "ANALYSE CLINIQUE" in prompt
+        assert "###RESULT###" in prompt
+
+    def test_prompt_das_type(self):
+        from src.medical.rag_search import _build_prompt
+
+        sources = [{"document": "cim10", "code": "E66", "page": 1, "extrait": "E66 Obésité"}]
+        contexte = {"sexe": "F", "age": 43}
+        prompt = _build_prompt("Obésité", sources, contexte, est_dp=False)
+
+        assert "DAS (diagnostic associé significatif)" in prompt
+
+    def test_prompt_enriched_context(self):
+        from src.medical.rag_search import _build_prompt
+
+        sources = [{"document": "cim10", "code": "K85", "page": 1, "extrait": "K85"}]
+        contexte = {
+            "sexe": "F",
+            "age": 43,
+            "imc": 34.4,
+            "duree_sejour": 6,
+            "antecedents": ["HTA", "diabète type 2"],
+            "biologie_cle": [("Lipasémie", "850", True), ("CRP", "45", True)],
+            "imagerie": [("TDM abdominal", "pancréatite stade C Balthazar")],
+            "complications": ["éruption cutanée"],
+            "dp_texte": "Pancréatite aiguë biliaire",
+        }
+        prompt = _build_prompt("Éruption cutanée", sources, contexte, est_dp=False)
+
+        assert "IMC 34.4" in prompt
+        assert "6 jours" in prompt
+        assert "HTA" in prompt
+        assert "Lipasémie" in prompt
+        assert "TDM abdominal" in prompt
+        assert "éruption cutanée" in prompt
+        assert "Pancréatite aiguë biliaire" in prompt
+
+
+class TestSearchSimilar:
+    """Tests pour search_similar avec score minimum et priorisation CIM-10."""
+
+    def test_filters_low_scores(self):
+        """Les résultats avec score < 0.3 sont éliminés."""
+        from src.medical.rag_search import search_similar
+        import numpy as np
+
+        mock_metadata = [
+            {"document": "cim10", "code": "K85", "page": 1, "extrait": "K85"},
+            {"document": "cim10", "code": "K86", "page": 2, "extrait": "K86"},
+        ]
+
+        mock_index = MagicMock()
+        mock_index.ntotal = 2
+        # Premier résultat score=0.9 (bon), second score=0.1 (sous le seuil)
+        mock_index.search.return_value = (
+            np.array([[0.9, 0.1]], dtype=np.float32),
+            np.array([[0, 1]], dtype=np.int64),
+        )
+
+        with patch("src.medical.rag_index.get_index", return_value=(mock_index, mock_metadata)), \
+             patch("src.medical.rag_search._get_embed_model") as mock_model:
+            mock_model.return_value.encode.return_value = np.array([[0.1] * 768], dtype=np.float32)
+            results = search_similar("pancréatite")
+
+        assert len(results) == 1
+        assert results[0]["code"] == "K85"
+
+    def test_prioritizes_cim10(self):
+        """Les sources CIM-10 sont priorisées (au moins 6 sur 10)."""
+        from src.medical.rag_search import search_similar
+        import numpy as np
+
+        # 8 sources CIM-10 + 8 sources guide_methodo, toutes avec bon score
+        mock_metadata = []
+        for i in range(8):
+            mock_metadata.append({"document": "cim10", "code": f"K8{i}", "page": i, "extrait": f"K8{i}"})
+        for i in range(8):
+            mock_metadata.append({"document": "guide_methodo", "page": i + 10, "extrait": f"Guide {i}"})
+
+        mock_index = MagicMock()
+        mock_index.ntotal = 16
+        scores = np.array([[0.9 - i * 0.03 for i in range(16)]], dtype=np.float32)
+        indices = np.array([list(range(16))], dtype=np.int64)
+        mock_index.search.return_value = (scores, indices)
+
+        with patch("src.medical.rag_index.get_index", return_value=(mock_index, mock_metadata)), \
+             patch("src.medical.rag_search._get_embed_model") as mock_model:
+            mock_model.return_value.encode.return_value = np.array([[0.1] * 768], dtype=np.float32)
+            results = search_similar("pancréatite", top_k=10)
+
+        cim10_count = sum(1 for r in results if r["document"] == "cim10")
+        assert cim10_count >= 6, f"Seulement {cim10_count} sources CIM-10 sur {len(results)}"
+
+
 class TestRAGSearchMocked:
     def test_search_similar_no_index(self):
         """search_similar retourne une liste vide si l'index n'existe pas."""
@@ -215,6 +422,7 @@ class TestRAGSearchMocked:
 
         assert diag.sources_rag == []
         assert diag.justification is None
+        assert diag.raisonnement is None
 
     def test_enrich_diagnostic_with_sources_no_ollama(self):
         """Enrichissement avec sources FAISS mais sans Ollama."""
@@ -238,11 +446,11 @@ class TestRAGSearchMocked:
         assert len(diag.sources_rag) == 1
         assert diag.sources_rag[0].document == "cim10"
         assert diag.sources_rag[0].code == "K85"
-        # Pas de justification (Ollama non disponible)
         assert diag.justification is None
+        assert diag.raisonnement is None
 
     def test_enrich_diagnostic_with_ollama(self):
-        """Enrichissement complet avec sources + Ollama."""
+        """Enrichissement complet avec sources + Ollama + raisonnement."""
         from src.medical.rag_search import enrich_diagnostic
 
         diag = Diagnostic(texte="Pancréatite aiguë biliaire")
@@ -259,6 +467,7 @@ class TestRAGSearchMocked:
             "code": "K85.1",
             "confidence": "high",
             "justification": "Pancréatite aiguë d'origine biliaire = K85.1",
+            "raisonnement": "1. ANALYSE CLINIQUE : La pancréatite...",
         }
 
         with patch("src.medical.rag_search.search_similar", return_value=mock_sources), \
@@ -268,4 +477,122 @@ class TestRAGSearchMocked:
         assert diag.cim10_suggestion == "K85.1"
         assert diag.cim10_confidence == "high"
         assert diag.justification == "Pancréatite aiguë d'origine biliaire = K85.1"
+        assert diag.raisonnement == "1. ANALYSE CLINIQUE : La pancréatite..."
         assert len(diag.sources_rag) == 1
+
+    def test_enrich_diagnostic_est_dp_flag(self):
+        """Le flag est_dp est bien passé à _build_prompt."""
+        from src.medical.rag_search import enrich_diagnostic
+
+        diag = Diagnostic(texte="Obésité")
+        mock_sources = [
+            {"document": "cim10", "page": 1, "code": "E66", "extrait": "E66 Obésité", "score": 0.9},
+        ]
+
+        with patch("src.medical.rag_search.search_similar", return_value=mock_sources), \
+             patch("src.medical.rag_search._call_ollama", return_value=None) as mock_ollama, \
+             patch("src.medical.rag_search._build_prompt", return_value="prompt") as mock_prompt:
+            enrich_diagnostic(diag, {"sexe": "F", "age": 43}, est_dp=False)
+            mock_prompt.assert_called_once_with("Obésité", mock_sources, {"sexe": "F", "age": 43}, est_dp=False)
+
+
+class TestEnrichDossier:
+    """Tests pour enrich_dossier avec le contexte enrichi."""
+
+    def test_enriched_context(self):
+        """enrich_dossier passe le contexte enrichi (bio, imagerie, etc.)."""
+        from src.medical.rag_search import enrich_dossier
+        from src.config import Sejour, BiologieCle, Imagerie
+
+        dossier = DossierMedical(
+            sejour=Sejour(sexe="F", age=43, duree_sejour=6, imc=34.4),
+            diagnostic_principal=Diagnostic(texte="Pancréatite aiguë biliaire"),
+            antecedents=["HTA", "diabète type 2"],
+            biologie_cle=[
+                BiologieCle(test="Lipasémie", valeur="850", anomalie=True),
+            ],
+            imagerie=[
+                Imagerie(type="TDM abdominal", conclusion="pancréatite stade C"),
+            ],
+            complications=["éruption cutanée"],
+        )
+
+        captured_contexts = []
+
+        def mock_enrich(diag, contexte, est_dp=True):
+            captured_contexts.append(contexte.copy())
+
+        with patch("src.medical.rag_search.enrich_diagnostic", side_effect=mock_enrich):
+            enrich_dossier(dossier)
+
+        assert len(captured_contexts) == 1  # DP seulement (pas de DAS)
+        ctx = captured_contexts[0]
+        assert ctx["sexe"] == "F"
+        assert ctx["age"] == 43
+        assert ctx["duree_sejour"] == 6
+        assert ctx["imc"] == 34.4
+        assert ctx["antecedents"] == ["HTA", "diabète type 2"]
+        assert ctx["biologie_cle"] == [("Lipasémie", "850", True)]
+        assert ctx["imagerie"] == [("TDM abdominal", "pancréatite stade C")]
+        assert ctx["complications"] == ["éruption cutanée"]
+
+    def test_das_gets_dp_context(self):
+        """Les DAS reçoivent le texte du DP dans leur contexte."""
+        from src.medical.rag_search import enrich_dossier
+
+        dossier = DossierMedical(
+            diagnostic_principal=Diagnostic(texte="Pancréatite aiguë biliaire"),
+            diagnostics_associes=[
+                Diagnostic(texte="Obésité"),
+            ],
+        )
+
+        captured = []
+
+        def mock_enrich(diag, contexte, est_dp=True):
+            captured.append({"texte": diag.texte, "est_dp": est_dp, "dp_texte": contexte.get("dp_texte")})
+
+        with patch("src.medical.rag_search.enrich_diagnostic", side_effect=mock_enrich):
+            enrich_dossier(dossier)
+
+        assert len(captured) == 2
+        # DP n'a pas dp_texte dans son contexte
+        assert captured[0]["est_dp"] is True
+        assert captured[0]["dp_texte"] is None
+        # DAS a dp_texte
+        assert captured[1]["est_dp"] is False
+        assert captured[1]["dp_texte"] == "Pancréatite aiguë biliaire"
+
+
+class TestFormatContexte:
+    """Tests pour _format_contexte."""
+
+    def test_minimal_context(self):
+        from src.medical.rag_search import _format_contexte
+
+        result = _format_contexte({})
+        assert result == "Non précisé"
+
+    def test_full_context(self):
+        from src.medical.rag_search import _format_contexte
+
+        ctx = {
+            "sexe": "F",
+            "age": 43,
+            "imc": 34.4,
+            "duree_sejour": 6,
+            "antecedents": ["HTA", "diabète type 2"],
+            "biologie_cle": [("Lipasémie", "850", True), ("CRP", "45", True)],
+            "imagerie": [("TDM abdominal", "pancréatite stade C Balthazar")],
+            "complications": ["éruption cutanée"],
+            "dp_texte": "Pancréatite aiguë biliaire",
+        }
+        result = _format_contexte(ctx)
+
+        assert "F, 43 ans, IMC 34.4" in result
+        assert "6 jours" in result
+        assert "HTA" in result
+        assert "Lipasémie 850" in result
+        assert "TDM abdominal" in result
+        assert "éruption cutanée" in result
+        assert "Pancréatite aiguë biliaire" in result